Development of Methods and Devices Towards Skin Elastography - Luca Bartolini by Luca Bartolini

Development of methods and devices towards skin elastography

Luca Bartolini

This thesis was reviewed by: prof.dr. Marloes L. Groot

VU University Amsterdam Amsterdam, The Netherlands

prof.dr. Gijs van Soest

Erasmus MC, Rotterdam, The Netherlands

prof.dr.ir. Theodoor H. Smit

Amsterdam UMC Amsterdam, The Netherlands

prof.dr. Ton G. van Leeuwen

Amsterdam UMC Amsterdam, The Netherlands

dr. Brendan F. Kennedy

The University of Western Australia Perth, Australia

ÂŠ 2020, Luca Bartolini. All rights reserved. No part of this thesis may be reproduced or transmitted in any form or by any means without permission from the author. Cover art by Chiara Verdoliva A digital version of this thesis is available at www.ubvu.vu.nl. The research in this thesis was performed in the Biophotonics and Medical Imaging section of the Department of Physics and Astronomy and LaserLaB at the VU University Amsterdam.

VRIJE UNIVERSITEIT

Development of methods and devices towards skin elastography ACADEMISCH PROEFSCHRIFT ter verkrijging van de graad Doctor of Philosophy aan de Vrije Universiteit Amsterdam, op gezag van de rector magnificus prof.dr. V. Subramaniam, in het openbaar te verdedigen ten overstaan van de promotiecommissie van de Faculteit der BĂ¨tawetenschappen op maandag 7 december 2020 om 9.45 uur in een online bijeenkomst van de universiteit, De Boelelaan 1105

door Luca Bartolini geboren te Cesena, ItaliĂŤ

promotoren:

prof.dr. D. Iannuzzi prof.dr. S. Gibbs

“All models are wrong, but some are useful” George Box

Contents

1 Introduction .......................................................................................................................... 15 1.1 Skin .................................................................................................................................................... 16 1.1.1 Skin â&#x20AC;&#x201C; why its biomechanics? .................................................................................. 16 1.1.2 Skin Anatomy .................................................................................................................. 17 1.2 Fundamentals of nanoindentation measurements ...................................................... 19 1.2.1 A brief history of indentation and nano-indentation .................................... 19 1.2.2 Principles of cantilever-based nanoindentation ............................................. 21 1.2.3 Terminology .................................................................................................................... 27 1.2.4 Specific mechanical models and common assumptions .............................. 30 1.3 Measuring skin with ferrule-top nano-indenters.......................................................... 36 1.3.1 Technological shortcomings .................................................................................... 37 1.3.2 The lessons we have learned â&#x20AC;&#x201C; Outline of the thesis ..................................... 38 1.4 A note on Optical Coherence Tomography ...................................................................... 40 1.4.1 Optical Coherence Elastography ............................................................................ 42 1.5 Outline of the thesis .................................................................................................................... 43 2 Comparison of frequency and strain-rate domain mechanical characterization ...................................................................................................................... 45 Abstract .................................................................................................................................................... 45 2.1 Introduction ................................................................................................................................... 46 2.2 Methods ........................................................................................................................................... 48 2.2.1 Sample preparation...................................................................................................... 48 2.2.2 Nano-indentation setup ............................................................................................. 48 2.2.3 Displacement-, load- and indentation-control mode .................................... 50 2.2.4 Dynamic mechanical analysis (DMA) ................................................................... 51 2.2.1 Nano-epsilon dot method (nano- ) .................................................................. 52 2.2.2 Comparing DMA and nanoresults................................................................. 55 2.3 Results .............................................................................................................................................. 56 2.3.1 Dynamic Mechanical Analysis (DMA) .................................................................. 56 2.3.1 Nano-epsilon dot (nano- ).................................................................................... 57 2.3.2 Comparing frequency and strain-rate domain results ................................. 58 2.4 Discussion ....................................................................................................................................... 60 2.5 Conclusion ...................................................................................................................................... 63

3 Multimodal probe for Optical Coherence Tomography epidetection and micron-scale indentation..................................................................................................... 65 Abstract: ....................................................................................................................................................65 3.1 Introduction: ..................................................................................................................................66 3.2 Experimental details:..................................................................................................................68 3.2.1 Ferrule-top sensor: design and fabrication........................................................68 3.2.2 Setup ....................................................................................................................................70 3.2.1 Experimental procedure .............................................................................................71 3.2.2 Sample preparation ......................................................................................................72 3.3 Results and discussion:..............................................................................................................73 3.4 Conclusions: ....................................................................................................................................76 4 Towards clinical elastography of dermal tissues: a medical device to probe skin's elasticity through suction, with subsurface imaging via Optical Coherence Tomography ....................................................................................................... 77 Abstract .....................................................................................................................................................77 4.1 Introduction ....................................................................................................................................78 4.2 Materials and Methods ..............................................................................................................80 4.2.1 Experimental Setup ......................................................................................................80 4.2.2 Sample preparation ......................................................................................................82 4.2.3 Data acquisition ..............................................................................................................83 4.2.4 Data analysis ....................................................................................................................84 4.3 Results ...............................................................................................................................................85 4.3.1 Static measurement on phantom, asymmetry factor ....................................85 4.3.2 Static measurement on skin ......................................................................................86 4.3.3 Dynamic measurement on skin ...............................................................................89 Measurement error......................................................................................................................89 4.4 Discussion ........................................................................................................................................90 4.5 Conclusions .....................................................................................................................................93 5 Valorization ........................................................................................................................... 95 5.1 Introduction ....................................................................................................................................96 5.1.1 Business Model Canvas ...............................................................................................96 5.1.2 Effectuation theory .......................................................................................................98 5.2 The four pillars ..............................................................................................................................99 5.2.1 Product-market fit .........................................................................................................99 5.2.2 Internal Analysis ......................................................................................................... 102 5.2.3 External Analysis ........................................................................................................ 105 5.2.4 Value Capturing ........................................................................................................... 111 5.3 A Summary, on Business Model Canvases ..................................................................... 112 5.3.1 Early Phase Business Canvas: from POC to TRL 5........................................ 112 5.3.2 Medium-Term Business-Model Canvas: reaching out ................................ 116 5.4 Conclusion .................................................................................................................................... 119

5.5 Disclaimer .....................................................................................................................................119 5.6 Chapter 5 - Appendix 1 ...........................................................................................................120 5.6.1 Phone Interview ..........................................................................................................120 5.6.2 Follow-up mail with form to fill in ......................................................................120 5.7 Chapter 5 - Appendix 2 ...........................................................................................................122 6 Conclusions..........................................................................................................................125 6.1 6.2 6.3 6.4

Viscoelasticity, virgin materials and control of sample, testing variables. ......126 Integration of subsurface imaging into ferrule-top force transducer ................126 Handheld elastometer with OCT imaging: .....................................................................128 Vaginal Probe ..............................................................................................................................130

7 Appendix 1 Excerpts from the â&#x20AC;&#x153;Investigative Medical Device Dossierâ&#x20AC;? .......131 7.1 Classification ................................................................................................................................132 7.2 Risk Management Plan ............................................................................................................134 7.3 Risk analysis ................................................................................................................................134 7.4 Risk evaluation ...........................................................................................................................136 7.5 Risk control ..................................................................................................................................137 7.6 Implementation of risk control measures ......................................................................139 7.7 Residual risk evaluation .........................................................................................................139 7.8 Risk-benefit analysis ................................................................................................................140 7.9 Risks arising from risk control measures .......................................................................140 7.10 Completeness of risk control ..............................................................................................140 7.11 Evaluation of overall residual risk acceptability .......................................................141 8 Summary ..............................................................................................................................143 9 Acknowledgements ..........................................................................................................145 10 References .........................................................................................................................151

List of Figures

Figure 1.1: Artistic representation of skin's anatomy.....................................................18 Figure 1.2: Vickers and Berkovich indentation imprints ..............................................20 Figure 1.3: Indenter setup and side-view of the ferrule-top sensor. ........................22 Figure 1.4: Region of contact between tip and sample ...................................................24 Figure 1.5: Indentation Curve, Load-vs-Displacement ...................................................25 Figure 1.6: Three-dimensional indentation curve ............................................................27 Figure 1.7: Terminology in mechanical testing..................................................................28 Figure 1.8: Finite Element Simulation of spherical indentation. ................................30 Figure 1.9: Oliver&Pharr model ...............................................................................................33 Figure 1.10: Samples ready for nanoindentation measurements..............................37 Figure 1.11: Healthy human skin, with a state-of-the-art OCT ...................................40 Figure 1.12: Comparison of Conventional OCT and OCE ...............................................43 Figure 2.1: Sample mechanical behaviour and properties ...........................................47 Figure 2.2: Dynamic mechanical analysis results on PDMS .........................................57 Figure 2.3: Apparent elastic moduli for virgin and pre-strained PDMS ..................59 Figure 2.4: Apparent moduli from nano- , and derived from DMA). ...................60 Figure 2.5: Nano-indentation setup ........................................................................................49 Figure 2.6: Prescribed indentation profiles for DMA and nano-ε M ̇ .........................53 Figure 3.1: Microscope view of the multimodal ferrule-top sensor..........................69 Figure 3.2: Multilmodal ferrule-top sensor fabrication steps......................................70 Figure 3.3: Overview of the setup. ...........................................................................................72 Figure 3.4: B-scan acquired with the multimodal sensor..............................................75 Figure 3.5: Load applied to reach 15μm indentation depth .........................................76 Figure 4.1: Medical device and enlargement of handheld probe ...............................83 Figure 4.2: Static measurement on the PDMS phantom.................................................87 Figure 4.3: Summary of all measurements on the PDMS skin-phantom ................88 Figure 4.4: B-scan of a volunteer’s left thumb fingertip.................................................87 Figure 4.5: C-scan of a volunteer’s left thumb fingertip. ................................................89 Figure 4.6: Dynamic analysis of the deformation .............................................................91

Figure 5.1: Building blocks of the Business Model Canvas........................................... 97 Figure 5.2: A visual metaphor for the Product-Market Fit. ........................................101 Figure 5.3: Market segmentation for Medical Devices in dermatology. ...............110 Figure 5.4: Day-One Business Model Canvas: ..................................................................115 Figure 5.5: Medium-term Business Model Canvas.........................................................117 Figure 5.6: Balance Sheet Forecast for the first 3 years. .............................................123 Figure 6.1: Range of frequencies accessible through our setup...............................127 Figure 6.2: OCE through a multimodal ferrule top ........................................................129 Figure 6.3: Concept design for transvaginal handheld OCE probe. ........................130 Figure 7.1: Failure modes and Effects Analysis (FMEA)..............................................142

1 Introduction

Chapter 1 — Introduction

1.1 Skin Billions of years ago in the oceans, Life is thought to have started inside a lipid membrane, that protected and enclosed increasingly complex chemical machinery. Nowadays, Life reveals itself in an astonishing number of forms, she roams on land, and she even questions its stance in the Universe. Yet – in her most tangible sense – she still is a wet laboratory enclosed in some sort of membrane. In the course of evolution, that membrane has become more specialized: in the animal kingdom it is an organ that we call skin. It is the largest human organ and its functions still include the original one of protection – against mechanical hazard, UV-rays, chemical, water-loss and microorganisms – and in combination with other systems, also serves functions of regulation and sensation (Agache & Humbert 2004). A discipline called Dermatology studies it, both in-itself, and as a general indicator of our physiology and health. Given the accessibility of skin, for ages its study relied on visual clues and touch. Modern advances of science, however, expanded the senses of dermatologists, who can now also count on tools to objectively investigate skin, accurately diagnose its conditions, and evaluate the effectiveness of therapies (Guy 1996). In this thesis, we will present the work we have done to explore a particularly representative aspect of skin: its mechanics. We will see how the initial approach to that research – microindentation – fell short of the purpose, and how we leveraged that situation to identify missing technologies and come up with the original solutions that constitute the backbone of this thesis. Specifically, in Chapter 2, we validate a time-dependent model for microindentation and in Chapter 3 we present a novel design to integrate depthresolved imaging into our microindenter. While those works contribute to the science of micro- and nanoindentation, we ultimately decided that our purpose required us to move to a radically different scale, one that could capture the behavior of skin as a tissue. In Chapter 4 we show the clinical-research device that we developed and built to investigate the mechanical properties of skin at the millimetric scale. Eventually, in Chapter 5 we present a business plan for the clinical translation of that research device.

1.1.1

Skin – why its biomechanics?

A typical example to intuitively grasp how the mechanical properties of a material can indicate its condition is the mundane act of choosing an avocado, judging its ripeness with a gentle nudge. Without us consciously realizing it, our brain is measuring the force we apply to the fruit and the induced displacement. In this

Skin â&#x20AC;&#x201D; 1.1

way we can estimate the elasticity and plasticity of the pulp, and finally judge the appeal of the avocado. This metaphor translates remarkably well for dermatologists, who still routinely use palpation (i.e. touch) to feel the stiffness of skin and accordingly establish their diagnoses (Cox 2006, 2007). The mechanical response of skin is an important indicator in a number of instances (Guy 1996, Elsner et al. 2001, Berardesca et al. 2014), including in the evaluation of therapeutic strategies (H. Dobrev 2000, Danin et al. 2012), in the study of the wound-healing process (Vogel 1970), as indicator of localized (Pakshir & Hinz 2018) or systemic (Enomoto et al. 1996, Eliceiri et al. 2014) diseases, in cosmetics (SALTER et al. 1993), and in biomedical applications (Leonardo 2010). It is therefore understandable that the interest in the formal, objective quantification of skinâ&#x20AC;&#x2122;s mechanical properties dates as far back as 1861, with the first studies of Karl Langer. After one-and-a-half centuries of continued effort, however, the objective mechanical characterization of skin is still an open problem.

1.1.2

Skin Anatomy

The functions skin serves, its structure, and its mechanics are deeply intertwined: to better understand dermal tissue and its mechanical behavior, we must first appreciate its anatomy and structure. Skin is part of the integumentary system, which includes derivatives such as hair, nails, and sebaceous and sweat glands. It makes up for a seventh of the body weight (Kanitakis 2002): such a vast organ, skin is adapted to the different regions it covers, from eyelids (so thin to be translucent in bright light) to foot soles (so leathery to resist running on gravel). The dominant feature of skin is its layered structure, where we distinguish three major layers: the epidermis, the dermis and the hypodermis (or subcutaneous tissue), as shown in Figure 1.1. The epidermis, enlarged in Figure 1.1B, is the outermost layer, standing as a first barrier to environmental hazards. It is avascular and comprised at 90% by tightly interconnected keratinocytes, a specialized type of cell containing a high concentration of filament proteins (keratins and others) that provide for its physical strength. The process through which the epidermis is generated is called cornification and its stages define the five layers in which epidermis is further subdivided.

Chapter 1 — Introduction

The life-cycle of keratinocytes (Cropley 2001) starts in the deepest layer of the epidermis, in a single layer of cuboidal cells called stratum basale, connected to the underlying dermis through the Dermal-Epidermal Junction. The cells in the stratum basale incessantly divide by mitosis, pushing their daughters towards the surface. During the migration, which takes about a month (Grover & Grewal 2008), keratinocytes undergo a differentiation process in which they create tight interconnection (desmosomes), then start producing copious amounts of keratin, progressively lose water and flatten, and ultimately die, forming the outermost, shedding layer called stratum corneum. It is the thickest layer in the epidermis – between 8 and 20 μm, but reaching up to 1.5mm on palms and soles – and it’s made of tightly packed layers of interlocking, fully-keratinized anucleate keratinocytes (Holbrook & Odland 1974). Due to these special characteristics it is sometimes considered an entirely separate layer from the underlying viableepidermis, with peculiar mechanical properties (Elias 2005, Geerligs 2010). The bulk of skin however, is found below the epidermis: it’s the dermis, in which special cells named fibroblasts secrete a dense fibrous connective tissue that makes up for the majority of its mass. The main components of that dermal matrix are collagen (a structural protein that represents 77% of the fat-free dry weight of the whole skin (Hussain et al. 2013)), elastic tissue (elastin and microfibrils proteins), and extrafibrillar matrix (an umbrella term that includes everything else, mainly the highly-hydrophilic glycosamino- and proteo-glycans) (Brown & Krishnamurthy 2019). Together with the fibroblasts that generate it, the matrix hosts a variety of other cells (mast cells and macrophages) and structures (called appendages), including capillaries from deeper arteries and veins, sebaceous and 18

Fundamentals of nanoindentation measurements — 1.2

sweat glands, hair follicles, nails, and nerve terminations. They all contribute to the various functions of the skin, and in creating a rich and complex structure underlying the apparently homogeneous epidermis. The constituents of the dermal matrix are primarily responsible for the macroscopic mechanical behavior of the dermis, and in turn of skin (Honari & Maibach 2014). Their mechanical response is what dermatologists perceive during palpation and it is heavily dependent on the loading condition. At small stresses, collagen fibers are randomly coiled and unable to bear loads. The elastic tissue is then responsible for the initial high compliance of the skin (Oxlund et al. 1988), the “softness” familiar to all of us. As stresses increases, the collagen fibers align and begin carrying load, considerably increasing the stiffness of the skin and providing it with its high strength (resistance to failure) and stiffness (resistance to deformation) that allow it to carry out its protective duties (Daly 1982). Skin also exhibits time-dependent mechanical properties, or viscoelasticity, associated with the supramolecular aggregates of collagen with ground substance (Bertassoni & Swain 2014). Below the dermis, and above the muscles, lays the hypodermis, also called subcutaneous fat tissue. It is mainly composed of adipocytes (fat storing cells) and connective tissues for thermal insulation and shock absorbance. It has been reported that even when applied loads are small, the hypodermis is relevant to skin mechanics (Diridollou et al. 1998).

1.2 Fundamentals of nanoindentation measurements Our initial approach to assess the mechanical properties of skin resembles the avocado test, though at a much smaller scale and using an optomechanical sensor in place of our finger. In simple words, we pushed a spherical tip onto the skin’s surface, measured and recorded the force applied and the displacement induced, to eventually characterize the observed mechanical behavior – knowing that it could hint to patho-physiological conditions. In the coming section, we will briefly explain the working principles of nanoindenters, introduce the physics used to obtain the mechanical properties, and finally look at the details of our measurement device.

1.2.1

A brief history of indentation and nano-indentation

Traditionally, “indentation” referred to hardness testing, with the first indentation device dating back to 1856 (Wade 1856). In those years, immediately following the Industrial Revolution, there emerged a need to control the quality metal 19

Chapter 1 — Introduction

batches in a non- or minimally-destructive way. Hardness testing was suitable for the purpose; it involved an indenter with a defined geometry (also called punch) pressed into a sample with a known load (in the simplest case, one can imagine a constant weight on the horizontal face of a sample). Upon removal of the punch, a small imprint remained in the material. It was measured by optical means – usually light microscopy – and used to numerically evaluate the hardness. That was the most inconvenient procedure in the process: the imprint was sometimes hard to find back at the microscope, and the determination of its size was the major source of error in hardness quantification. Nevertheless, it was a necessary step and remained unchanged for decades. Among the most widespread protocols, still in use, are Brinell, with spherical indenters, and Vickers, with a diamond pyramid (Walley 2012) – which can be seen in Figure 1.2. More than a century later, the 1980s saw the introduction of an extremely versatile technique: atomic force microscopy (AFM) (Binnig et al. 1986). Applications were found in many fields, including in indentation, where AFM contributed to a major paradigm shift towards the nano scale. The accessible scales were unprecedentedly small, with resolution of ~1 micronewton for forces and ~0.2 nanometer for displacements. Remarkably, AFM allowed for the simultaneous measurement of the history of applied forces and induced sample displacement for the whole duration of a loading cycle, and could therefore leverage depth-sensing techniques that were previously developed.

Figure 1.2: A) Top-view, Imprint left by a Vickers pyramid indenter on hardened steel. (Clarke 1988). B), C) and D) Meridial-plane view of spherical indentation, respectively on plasticine, mild-steel, and interpretation (M. C. Shaw & deSalvo 1970)

Fundamentals of nanoindentation measurements — 1.2

Macroscopic depth-sensing, or instrumented indentation, existed from long before, but the comparably complex and expensive setups limited its application, so that the general interest from the engineering community was limited. Moreover, the first prototypes of depth-sensing hardness-indenter were from the Soviet Union, and “[with respect to English-speaking scientist,] the activity of their colleagues from the former Soviet Union is less well known because the results of Soviet researchers were mainly published in Russian.”(Borodich 2014) Therefore, even if the origin of Contact mechanics1 is attributed to the work of Heinrich Hertz in 1882 (Hertz 1882), as put by Johnson(Johnson 1987) “developments in the theory did not appear in the literature until the beginning of the century, stimulated by engineering developments on the railways, in marine reduction gears and in the rolling contact bearing industry”. Even later, the newfound application of instrumented-indentation techniques via AFM resulted in a blooming field, that popularized the mechanical characterization by nanoindentation, which has now been applied to all types of materials, from metals, ceramics, bone tissue, to polymers and soft-biological samples, down to cells or even single vesicles with a diameter of 100nm (VanLandingham 2003, Lin & Horkay 2008).

1.2.2

Principles of cantilever-based nanoindentation

An AFM nanoindenter is essentially a force transducer in the form of a cantilever beam, of which one end is fixed, and the other is hanging free and equipped with a tip of known geometry. A piezoelectric element controls the vertical translation of the whole sensor towards the sample. Following contact, the tip is pushed into it, and the cantilever bends to balance the sample’s reaction force. The position of the tip is continuously tracked by triangulating a laser beam reflected off the surface of the cantilever, whose deflection with respect to the equilibrium position we call . Under small deflections2, we can approximate the cantilever to a linear spring, whose spring constant can be obtained through several techniques (Burnham et al. 2003, Clifford & Seah 2009). Then, the load

applied by the tip onto the sample is given by Hooke’s law: =

⋅

(1.1)

The branch of solid mechanics that describes deformations and forces that occur when two bodies come in contact.

As laid out in Euler’s beam theory, when a transversal load is applied at the free end of the cantilever, such that the deflection is much smaller that its length, it is possible to approximate the cantilever’s behavior to that of a linear spring.

Chapter 1 — Introduction

Equation 1.1 links a direct, optical measurement (the deflection of the cantilever) to the restoring force of the cantilever-spring, that is also the Load applied to the sample: one of the two quantities of interest in nanoindentation experiments (we will see the other, indentation depth, later in this section). In this work, we use ferrule-top sensors, which also are cantilever-based force transducers, with the fundamental difference that the deflection tracking is carried out by means of Fabry-Pérot interferometry via an optical fiber aligned with the free hanging end of the cantilever.(D. Chavan et al. 2010, 2012, D Chavan 2014) Ferrule-top sensors are all-optical: the signal travels on a fiber to-and-from the distal electronics. With respect to AFM, that design results in a sturdier sensor, with a simpler calibration procedure and which can easily operate in harsh environments. The technology is patented (Iannuzzi et al. 2011), it has found commercial viability, and has been used in more than a hundred peer-reviewed studies. A schematic view of the ferrule-top indenter and a microscope image of the sensor are shown in Figure 1.3.

Figure 1.3: A) Schematics of the indenter setup. A C-shape frame hosts an X-Y stage at the bottom, holding the sample; a coarse-approach actuator allows the ferrule-top sensor to reach the proximity of the sample’s surface. The piezoelectric element drives the actual indentation B) Annotated microscope side-view of the ferrule-top sensor. The Fabry-Pérot cavity is between the cleaved end of the detection fiber and the reflective upper surface of the cantilever. Note that at equilibrium, dcantilever≡0

Fundamentals of nanoindentation measurements — 1.2

The signal from the cavity is demodulated and linearized by a commercial interferometer, which offers a direct measurement of the change of the FabryPérot cavity length, equal (and opposite in sign) to . During indentation experiments, the load applied to the sample is calculated by multiplying by the spring constant of the cantilever, according to Eq. 1.1. In order to obtain the second quantity necessary for mechanical characterization, namely the indentation depth, we make use of a geometrical relation between the vertical displacements at play. The ferrule-top is rigidly fixed to a piezoelectric element that controls the vertical translation of the whole sensor. The piezoelectric element, is in turn rigidly fixed on the rigid frame of the indenter.3 Then, when the sensor is in contact with the sample, the extension of the piezoelectric element (i.e. how much the whole ferrule is translated towards the sample) is divided among the deflection of the cantilever , and the indentation depth (also sample displacement) , as summarized in eq 1.2: =

(1.2)

is prescribed and known at all times, and the cantilever Considering that bending is measured by the interferometer, Eq. 1.2 provides us with a straightforward way to obtain the indentation depth , that together with the applied load, will later be used to calculate the mechanical properties. Figure 1.4 shows a schematic of the contact region with the most important variables at play. In such an indenter, the only variable that we can directly act on . Its subsequent partition into and is the piezoelectric extension cannot be known a-priori since it depends on both the (known) stiffness of the cantilever and the unknown stiffness of the sample: we can only measure it. However, it is highly desirable to reach a predetermined indentation depth, or apply a calibrated load on the sample, regardless of the unknown stiffness of the material. With this aim, feedback loops have been implemented so that an arbitrary history of load, or indentation depth can be defined. During the experiment, the loop will correct to impose the desired setpoints of load or indentation depth.

Here “rigid” means that deformations in the ferrule body or in the indenter frame are negligible. The load applied to the sample is indeed routed along the C-shape frame that holds it from the bottom, therefore a deformation will be induced in the frame or in the body of the ferrule. However, the forces at play are in the μN range, a magnitude too low to induce appreciable deformations in the frame of the indenter or in the glass body of the ferrule sensor.

Chapter 1 — Introduction

1.2.2.1

The raw-data

During a nanoindentation measurement, the evolution of the load and sample displacement are simultaneously recorded at high frequency. To provide the reader with tangible numbers, in the experiments that will be presented in this thesis, typical loading cycles last in the range of seconds, at acquisition frequencies in the range of (0.5-16) kHz, so that several thousands of data points are recorded in a single measurement, each with indentation depth, applied force, and their timestamp. Figure 1.5 shows an example of indentation curve, plotting the two quantities that we obtain with equations 1.1 and 1.2, respectively the load on the sample, and the indentation depth. The measurement was acquired with piezo-control (load and indentation depth are not user-defined) on a fibroblast culture, i.e. a 1mm thick collagen scaffold that had been seeded with fibroblasts and incubated for three weeks prior to the measurement. The measurements were done in liquid (physiological solution) to keep the culture in a suitable medium and minimize the capillary forces responsible for tip-sample adhesion.

Figure 1.4: Close-up onto the region of contact between tip and sample. The annotated quantities are the Indentation Depth (or Material Displacement) δ, the Load P, the Radius of the tip R. Note that even if the “real” contact area is a spherical cap, in the assumption of normal load (see section 1.2.4), will only be considered as such its horizontal projection, i.e. the base of that cap with radius a. The Z axis is also referred as “axial” direction, whereas X is “transversal”.

Fundamentals of nanoindentation measurements — 1.2

Figure 1.5: Indentation Curve, Load-vs-Displacement, on a culture of fibroblasts. Salient points and regions of the curve are annotated. The red line indicates the initial Hertzian-fit, the deviation at increased indentation depth is ascribable to viscoelastic effects.

In a typical indentation curve, we can identify a few salient features: • At the contact point the tip reaches the surface of the sample, the load starts increasing, and δ is set to 0. • The initial portion of the curve shows a Hertzian Fit (see section 1.2.4.1), which, as the indentation proceed and viscoelastic effects subside, becomes less and less appropriate. • The forward stroke, or loading, (in blue) identifies the region there the load increases and the tip moves deeper into the sample • The hold portion (in yellow) is at a slope • The retraction portion of the curve, or unloading, as the word implies, is measured when the piezo is retracted, until it gets out of contact. Note that while the curve in Figure 1.5 is the most common way to present indentation data, it does not directly show time. The full representation of the raw data is in 3D: indentation, load and time, as shown in Figure 1.6. Figure 1.6 shows what happens during the hold phase, and consequently, the importance of feedback controls. In the measurement that provided the data, after the piezoelectric element drove the forward stroke, its extension was kept 25

Chapter 1 — Introduction

constant for a “hold time”. In that phase, viscoelastic phenomena clearly manifest in a mix of stress-relaxation (decreasing load) and creep (increasing depth), which we will discuss further in section 1.2.4.3. The mathematical formulations to characterize stress-relaxation or creep, however, respectively require either a constant indentation depth or load, condition that was not realized at the time of the experiment. For viscoelastic materials, this is a severe shortcoming, and it remarks the importance of indentation or load control loops, which would allow a user to predefine the history of Load or of indentation depth to realize during an experiment. 1.2.2.1

The determination of the contact point

A crucial aspect of experimental nanoindentation is the determination of the contact point , i.e. when the tip has reached the sample’s surface and we set = 0. Since the sensor starts out of contact, and the position of the surface is not known, that point cannot be known a priori. A good amount of literature deals with this issue (Benítez et al. 2013, Gavara 2016), and several ways to obtain it are available, both during an experiment and in postprocessing. Here, we briefly present our strategies for the contact point determination. In each measurement, the sensor initially hovers over the sample, out of contact. It is then moved towards the surface, while the deflection of the cantilever is continuously recorded. While out of contact, excluding small inertial effects and noise, the cantilever bending will be compatible with 0 (no bending). The condition to set the contact-point is triggered as the cantilever deflects more than a predefined threshold, usually 3 times the out-of-contact noise level. This method slightly and intrinsically overestimates the contact point, but it is necessary to identify the domain of validity of Eq. 1.2, that we use in the control loops, to impose a user-defined history of loads or displacements. The contact-point determination in post-processing offers better accuracy, and we carried it out by means of a piecewise fit of the measured load against the indentation depth: =0 = ƒ( )

< ≥

where ƒ( ) is the model of choice that relates load to indentation depth (and any other variable), in our case the Hertzian model (see following subsection 1.2.4.1), and the fit contains as a free parameter.

Fundamentals of nanoindentation measurements — 1.2

Figure 1.6: Three-dimensional indentation curve, same data from Figure 1.5, with the addition of the “time” dimension. The color intensity on the 2D projections maps the density of projected points.

1.2.3

Terminology

The mechanical characterization is completed by fitting the measured data to the chosen analytical model from contact-mechanics, to finally obtain the optimal parameters that describe the observed measurement, i.e. the “mechanical properties” of the sample. Before taking a closer look at specific mechanical models, it is important to clarify some of the terminology, since in the field of continuum mechanics, some commonly used words have a special meaning. To do so, we will go through the process of a measurement and explain its most important aspects with the perspective laid out by Mattei et al. (Giorgio Mattei & Ahluwalia 2016) and summarized in Figure 1.7. The first distinction is between mechanical behavior and mechanical properties. The former is the name given to the directly-observable, intrinsic materialresponse, affected by all the so called “sample variables”, including its composition, source, preparation and preservation status. 27

Chapter 1 — Introduction

Figure 1.7: Terminology in mechanical testing – concepts adapted from (Giorgio Mattei & Ahluwalia 2016).

Every mechanical measurement begins with the imposition of a mechanical stimulus, be it a tensile load, a compressive, bending or an indentation (the testing method); in our case that is a calibrated load or indentation profile. During the stimulus, the elicited mechanical response, i.e. the mechanical behavior, is continuously measured. The observed mechanical behavior is therefore as unique and peculiar as the specific sample under investigation. Scientists, however, are often interested in broader-encompassing qualities than those of the tangible sample in front of them. They are interested in the mechanical properties: the “numbers” that characterize and summarize the mechanics of our sample, allow for comparisons, and offer diagnostic power; those that we could use to “quantify” the ripeness of our avocado. Those mechanical properties are rigorously defined within a discipline called Continuum Mechanics, where they are the normalized quantities that model the relation between stress, the internal distribution of forces, and strain, the internal distribution of displacements. In their full representation, stress and strain are 2nd-order tensors, whose mathematical manipulation – and physical measurement – is involved and often out of reach: in experimental mechanics, it is necessary to find ways to reduce the complexity of tensorial stress and strain to vectors or even scalars, resorting to assumptions of symmetry. The simplest, and most widely known example is the case of uniaxial load, where a force is applied to the surface of a body of length !. “Engineering stress and strain”, respectively " and in Eq. 1.3, are then simplified as scalars, defining stress as a pressure (force over cross-sectional area ), and strain as a relative change in length:

Fundamentals of nanoindentation measurements — 1.2

$! !

(1.3)

Mechanical models are constitutive equations, i.e. parametrized relationships between stress and strain, whose parameters we call mechanical properties. Models take specific forms under specific assumptions; for example, in the simplest case – a linear, purely-elastic material – we can write Hooke’s law4 to relate stress and strain through a single parameter, the elasticity modulus (Young’s Modulus) %: "=%⋅

(1.4)

In Eq. 1.4, % is the Mechanical Property, (or descriptor): a number – the only one in this case – needed to fully summarize and characterize the sample’s idealized mechanical behavior. In nanoindentation, however, stress and strain are not scalars: their spatial distribution is much more inhomogeneous. Figure 1.8 shows a simulation that calculates the internal Von Mises stress resulting from spherical indentation on a purely elastic material, which illustrates how the load and displacement measured at the surface are a mere window on the internal distributions of stress and strain. While it is beyond the scope of this Chapter to dig into the subtleties of this matter, it is worth reminding that that first conceptual step to obtain any mechanical property, often hidden inside the formulas, is to transform the observed load and displacement into “digestible” forms of stress and strain, that are consequently used in mechanical models.

In case of a three-dimensional linear elastic material, we would need to use the tensorial form of the relationship, called Generalized Hooke’s law.

Chapter 1 â&#x20AC;&#x201D; Introduction

Figure 1.8: A) Finite Element Simulation of spherical indentation, displaying Von Mises stress and resulting deformation (Baniasadi et al. 2015).

1.2.4

Specific mechanical models and common assumptions

There is a vast literature of more sophisticated material-models, whose complexity varies on a spectrum. For instance, it is possible to include in the analysis time-dependency (stress- and strain-rates), material structure (e.g. layers), poro-elasticity (fluid-flow inside solids), or other phenomena (e.g. adhesion, friction). The description of any real material generally requires only a subset of those assumptions. For example, the study of metals or ceramics must keep into account elasticity and plasticity (irreversible deformations), while it can often ignore the rapid transient dynamics. On the other hand, for most polymers and biomaterials, plasticity is less relevant, while time-dependent behavior, i.e. viscoelasticity is remarkably insightful. We must keep in mind that every material model will inevitably imply some sort of approximation â&#x20AC;&#x201C; assumptions that simplify the mechanical behavior so that we can capture, simplify and describe it in a meaningful way. Therefore, the scientist needs to carefully consider the tradeoff between the usefulness of a concise model with a few, intuitive, descriptors, and the accuracy of more complicated ones with several tunable parameters â&#x20AC;&#x201C; a balance between oversimplification and overparametrization.

Fundamentals of nanoindentation measurements — 1.2

In the coming section, we will look more carefully at the importance of choosing the right model, which in turn is the reflection of acceptable assumptions. All the following assumptions are shared by the models that will be discussed: 1.

4. 5. 6. 7.

The material of the sample is isotropic (its properties are direction independent), linear (its properties do not depend on depth) and homogeneous (it has no internal structure) The sample is half-infinite: its physical dimensions are large enough with respect to the indentation depth, that boundary effects (either at the sides or at the bottom of the sample) can be neglected. As a rule-of-thumb, this is guaranteed when the sample’s thickness and width is at least 10 times the maximum reached The sample-tip interaction is one of “normal contact” as defined in Contact mechanics, i.e. the applied load is normal to the surface, and the adhesion and the friction between indenter and sample are neglected The surface roughness is negligible with respect to the radius of the tip The contact area is a function of the indentation depth and of the tip radius The tip of the indenter is spherical, and much stiffer than the sample The indentation depth is small with respect to the radius of the tip. As a rule of thumb, the condition is considered acceptable for < &/10

1.2.4.1

Hertz

Long before depth-sensing indentation was conceived, Hertz (Hertz 1882) published a work about the contact of elastic bodies. His work was general and applied to purely-elastic bodies with varying symmetries and curvatures: to this day, it is used in engineering to estimate forces at play in two-body contact, for example in gears or ball bearings. Spherical indentation is a special case of his theory, one that found an incredibly vast field of applications. It postulates that the force required to indent an elastic sphere of radius & to a depth takes the following form: =

4 ∗ -/ % & . 3

0/ .

(1.5)

where % ∗ is the reduced Young’s Modulus, which in the approximation of rigid tip (common assumption #6) reduces to: %∗ =

%1 2 1 − 41. 2

(1.6)

where 41 2 , the Poisson’s coefficient, is generally accepted to be 0.5 (incompressibility) for biomaterials (Fung & Skalak 1982, Humphrey 2003), with 31

Chapter 1 — Introduction

the awareness that, was the real value slightly different, only a systematic offset would be introduced in the estimation of the elasticity. Then, the only free parameter in Eq. 1.6 is the Elastic modulus of the sample %1 2 , which we can obtain by a fit on the raw data of load and indentation depth , as shown in Figure 1.5. It is a mechanical property: a number that describes the intrinsic behavior of the sample, and quantifies its elasticity. While the model strictly holds true only for purely elastic materials, it can be applied also in the initial loading phase of viscoelastic materials, such as the softand biomaterials that we investigated: for a short time after the contact, viscoelastic (time-dependent) effects can be neglected. 1.2.4.2

Oliver Pharr model

One of the first models specifically developed for depth-sensing nanoindentation in modern times is the Oliver-Pharr model (Oliver & Pharr 1992, 2004). It addressed the engineering need to reliably measure the hardness5 of stiff nonbiological samples, which contributed to its widespread success: the first paper has, to this day, more than 22-thousand citations. The model is applicable to elastic-plastic materials, capturing the two most prominent behaviors of metals and ceramics. Elasticity is the ability to return to the initial shape after a deformation. Plasticity is the phenomenon that causes the irrecoverable change in shape that occurs when the deformation overcomes a certain threshold. Plasticity is the reason for which the indenter leaves a permanent imprint, so that during the unloading the tip gets out of contact at a depth 5 , as shown in Figure 1.9 (note that this holds true only if viscoelasticity is neglected, as in the model under discussion). In the Oliver-Pharr derivation, it is assumed that the loading portion is purely elasto-plastic, whereas the unloading is purely elastic (Borodich 2014), so that the Hertzian model can be used in that region. With this approach, the loading takes the following form: =

4 ∗ -/ % & .( − 3

0/ 5) .

(1.7)

5 “The deformation can be produced by different mechanisms, like indentation, scratching, cutting, mechanical wear, or bending. In metals, ceramics, and most of polymers, the hardness is related to the plastic deformation of the surface. Hardness has also a close relation to other mechanical properties like strength, ductility, and fatigue resistance, and therefore, hardness testing can be used in the industry as a simple, fast, and relatively cheap material quality control” (Broitman 2017).

Fundamentals of nanoindentation measurements — 1.2

In Eq. 1.7, 5 is the depth of the remaining imprint, subtracted from , the indentation depth, so that the load 6 = 5 7 = 0. In other words, the focus shifts to the unloading portion of the curve, that starts as the indenter reaches its maximum depth and retracts and ends at the imprint depth.

Figure 1.9: Oliver&Pharr model arguably owes part of its success to the simple features of the indentation curve that are needed to calculate the Young’s Modulus.

The model also proposed an even simpler approach to extract the Elasticity modulus, and its derivation is well explained in (Kontomaris & Malamou 2020). By differentiating with respect to , Young’s modulus is obtainable by: %=

(1 − 41. 2 29&

2 :

)

;

(1.8)

where ;, the contact stiffness, is the initial slope of the unloading portion of the experimental ( ) curve. 1.2.4.3

Viscoelastic Modeling: DMA – Dynamical Mechanical Analysis

Soft- and biomaterials exhibits a remarkably time-dependent mechanical behavior: they are viscoelastic. The aforementioned Hertz and Oliver-Pharr models does not include this important feature; in this subsection we will show our chosen approach to characterize it.

Chapter 1 — Introduction

Viscoelasticity is most evident in experiments of stress-relaxation and creep. In stress-relaxation, a constant displacement (strain) is applied to the body; as time passes, the load (stress) required to remain at that displacement decays (relax) up to an equilibrium value. Creep experiments are analogous, in that a constant load is applied to a body, and the resulting displacement is observed to increase (creep) in time. While it is possible to measure the viscoelasticity of a material through nanoindentation creep or stress-relaxation (Michelle L Oyen 2005), there are advantages gained by moving from those time-domain measurements to frequency-domain ones, including on noise sensitivity and data processing. In particular, we implement Dynamic Mechanical Analysis (DMA) (van Hoorn et al. 2016), an oscillatory nanoindentation technique originally proposed by Herbert and (the “familiar”) Oliver and Pharr (Herbert et al. 2008). DMA is based on the decomposition of the mechanical response into its elastic and viscous responses. The elastic component, quantified by the storage modulus %’, is the material’s ability to store energy. Conversely, the viscous component is related to the dissipative processes, and is characterized by the loss modulus %’’ (Gutierrez-Lemini 2014). In DMA, the storage and loss moduli are measured at a discrete set of frequencies, in our case, usually, logarithmically spaced frequencies in the range accessible by our measuring setup, i.e. ≈0.01-10Hz. At each frequency, the load is oscillated and the corresponding displacement is measured6. In any nanoindentation setup it is not possible to apply negative loads (aside from friction and adhesion phenomena), so the oscillation is performed around an offset 1 , with an amplitude smaller than the offset.

Provided that the is small enough to keep the response within a local, linearviscoelastic region(G. Mattei et al. 2014), the measured displacement will also be sinusoidal and will present a phase lag = that increases with the viscous component. While a purely elastic material would present no phase-lag and a purely viscous would present a 90° one; a real material exhibits a value in between the two extremes. The storage and loss moduli at each probed frequency are then obtained through the following set of equations:

It is equally possible to impose a displacement, and measure the resulting reaction force.

Fundamentals of nanoindentation measurements — 1.2

⎧ %′ = ⎪1 − 4 . ⎨ %′′ ⎪1 − 4. = ⎩

1 F 1 sin(=) F

cos(=)

(1.9)

and are respectively the amplitude of the imposed load and where measured displacement, and F is the radius of the contact area at the offset load, calculated at the depth ℎ1 reached at the offset load, so that F = 2 9ℎ1 &.

Therefore, DMA does not return a single mechanical property, as Hertz and Oliver&Pharr model do, but a pair of storage and loss moduli for each probed frequency. Figure 2.4 is an example of the results of a typical DMA experiment. 1.2.4.4

(nano)Epsilon Dot – Time-domain measurements on virgin materials

The oscillations in DMA measurements must take place around a static pre-load 1 oscillated at a depth δ1 , so that the tip stays in contact during the whole cycle. However, one might want to investigate the mechanical properties of a virgin material, i.e. perform a measurement without any pre-load or pre-stress. Moreover, if the assumption of strictly linear material is dropped, there might be interest in measuring the properties around δ = 0, which is impossible with DMA.

The epsilon-dot method allows for measurements in that range, by probing the sample at different strain rates and measuring its apparent stiffness. A simple way to intuitively grasp this method is imagining the different response obtained when stretching dough at different speeds: a gentle pull would result in little, flow-like resistance, whereas a quick jerk could be opposed with much higher stiffness.

This strain-rate dependence is characteristic of all viscoelastic materials, and the epsilon-dot method aims at capturing it quantitatively, on a virgin material. Chapter 2 elaborates on the equivalence of testing the viscoelastic properties in the frequency- and in the time-domain, comparing results from epsilon-dot and from DMA. 1.2.4.5

Beyond our assumptions

The seven assumptions listed in section 1.2.4 allow us to simplify the indentation problem and write down analytical relationships between load and displacement. It would also be possible to release some of those at the cost of a more involved mathematical analysis. Above all, it is worth mentioning models to investigate adhesion and non-linearity.

Chapter 1 — Introduction

Adhesion is a surface-dependent effect that overlaps with the volumedependent bulk mechanical response probed by indentation. Therefore, the effect is larger at smaller scales, and in nanoindentation is often non-negligible (unless mitigated by measuring in a liquid environment). The JKR model(Johnson et al. 1971), named after Johnson, Kendall, and Roberts, is the most widespread approach in the regime defined by nanoindenters; however, “nowadays there are several well-established classic models of adhesive contact, which include not only the JKR theory, but also the DMT theory and the Maugis transition solution between the JKR and DMT theories” (Borodich 2014). Non-linearity is commonly addressed by the so called “Continuous Stiffness Measurements” (X. Li & Bhushan 2002) based on oscillatory measurements at various depths, at each of which the amplitude is small enough that the material response can be considered linear, so that simpler model can be applied piecewise. Finally, radically different and powerful is the approach granted by (inverse) Finite Element Methods. The direct version of the technique, FEM, allows to run mechanical simulations with arbitrary material properties, structures and indenter shape. Inverse FEM (iFEM) iteratively runs FEM simulations, looking for the optimal material properties that match experimentally acquired loaddisplacement curves(Kauer et al. 2001, Lei & Szeri 2007, Yao et al. 2014). While iFEM can be extremely powerful, its disadvantages include its high sensitivity to boundary conditions, and the time-intensive calculations required.

1.3 Measuring skin with ferrule-top nano-indenters Our journey to investigate the mechanical behavior of skin started with the Piuma, a commercial nanoindenter based on ferrule-top technology. It was one of the first versions of the instrument, and our first objective was to develop a reliable measurement and data analysis protocol. In the rare occasions they were available, we measured excised scars or healthy tissue from abdominoplasties, and at other times, fibroblasts cultures. The samples in those first experiments are shown in Figure 1.10. In what perhaps is a common course of events, those initial measurements were entirely unfruitful. Measurement after measurement, it became clear that even if the device was working as intended, we were out of its domain of applicability: what we had set out to measure was out of reach.

Measuring skin with ferrule-top nano-indenters — 1.3

Therefore, in the second year of my PhD project, we took a step back, paying attention to previous shortcomings and the challenges we were facing, and we set out to develop new and improved tools to investigate skin.

Figure 1.10: Samples on the holder, ready for nanoindentation measurements with the commercial nanoindenter. A) Human keloid scar B) cellularized collagen scaffold (i.e. fibroblast culture) in physiological solution.

1.3.1

Technological shortcomings

The issues in those first measurements were of two types: intrinsic technological limitations of the tool we were using, which were later solved, and those that created room-for-improvement and drove the work of the following years. The first technical limitation was the 20μm excursion of the piezoelectric element responsible for the vertical translation of the sensor. While it is a typical specification, it was insufficient to meaningfully measure real, macroscopic samples, especially skin, with asperities and irregularities that can reach 200μm for the deepest furrows (Geerligs 2010). In particular, the first version of the Piuma worked with a two-stage system on the vertical axis. The first stage was a low-resolution (>1μm) stepper motor that provided the long travel (>50 mm) necessary to approach the sample and stow the sensor between measurements. The second stage, mounted on the first and carrying the ferrule-top sensor, was the actual piezoelectric element with a 20μm excursion, that reached the accuracy and resolution necessary to perform a nanoindentation experiment. By looking at Eq. 1.2, it understandable that to obtain any reasonable indentation depth (e.g.

Chapter 1 — Introduction

more than a couple of μm)7, the sensor must already be within a few μm from the surface before the recording starts. This was ensured by a find-surface algorithm, in which the stepper was moved in steps of 7μm until the cantilever detected a bending, indicating contact with the sample. The sensor was then retracted, and a piezo-controlled measurement started. The find-surface procedure resulted in a small pre-stress on the material, but most importantly, due to the high noise introduced by the stepper motor, was prone to a failure mode that would jam the probe into the sample, and another one that would not retract the sensor enough to start the measurement when out of contact. Uncountable measurements had to be discarded in post-processing and many broken sensors had to be replaced. The second technological shortcoming was the lack of feedback control: it was not possible for the user to prescribe a load, or an indentation depth, but only the extension of the piezoelectric element. In an experiment, upon contact and following Eq. 1.2, that extension would be divided according to both the stiffness of the cantilever, and the unknown mechanical behavior. This indetermination resulted in the impossibility to test skin at a constant depth. Since the outer layers of skin are highly stratified (see section 1.1.2), we observe highly depthdependent mechanical properties: the inability to control for the measurement depth was detrimental. Well after we decided to follow other research directions, all our ferrule-top nanoindenters – commercial and research ones – included solutions to both aforementioned shortcomings. The first was solved by using long-travel piezoelectric elements, that offer an excursion of at least 300μm and solve the problem of surface approach at its root. The second was addressed by feedback loops, that allow us to directly control (and prescribe) the evolution of either indentation depth or the applied load. We now recognize those solutions as a pillar of successful nanoindentation on viscoelastic materials.

1.3.2

The lessons we have learned – Outline of the thesis

The thread that connects all the works in this thesis is in how we approached the following issues, namely the model we used, the peculiarities of the sample, and, encompassing all of the above, the measurement scale.

For example: we could imagine a sensor hovering 4μm above the surface, with a stiffness comparable to the one of the sample. From the maximum extension of the piezo (20μm), 4μm would go into the surface approach, and assuming similar cantilever bending and indentation depth, would result in each being only 8μm.

Measuring skin with ferrule-top nano-indenters — 1.3

In 1966 Psychology of Science (Maslow 1966), Maslow talks about the “law of the instrument”, a cognitive bias summarized in the maxim “if all you have is a hammer, everything looks like a nail”, which brilliantly encapsulates one aspect of the first measurements. Initially, the Piuma used to provide only the Young’s Modulus as based on the most widespread model in nanoindentation: Oliver Pharr. As we’ve seen in the previous section, however, the model was developed for elasto-plastic materials – of which biomaterials and skin are not an example of – and it neglected any time-dependent behavior, which we now know is a fundamental feature of skin’s mechanical characterization. Once we realized its importance, we started looking into viscoelastic models – above all, DMA and epsilon-dot method – which eventually, resulted in the paper presented in Chapter 2. The usefulness of the mechanical models presented so far comes from their simplicity, which is based on assumptions of homogenous and isotropic materials. However, skin is a really complex tissue, layered, and with plenty of different inner structures, and it proved to hardly fit within the boundaries set by the approximations and assumptions of those simple models. We realized that it would be extremely useful to be able to visualize the internal structure of skin before indentation, and that the benefit would also apply to all samples that cannot be considered homogeneous. So, we studied a novel design to integrate an imaging technique, namely Optical Coherence Tomography, in our nanoindenter. That work is presented in Chapter 3 Eventually, we realized that what we wanted to measure – the pathophysiology of skin – was better accessible at a larger scale than that affordable by a nanoindenter. With the latter, we were at most reaching depths of ten-to-twenty microns: comparable to the thickness of stratum corneum (section 1.1.2). Yet, while there are many studies that are interested in its mechanical characterization, the stratum corneum is made of dead keratinocytes, therefore its diagnostic power is very limited. To investigate the mechanics of skin as a tissue, we needed a technique that would let us access the mechanical behavior of the dermis, and that was only possible with a radically different tool, meant to produce a mechanical stimulus that would reach millimeters depth. Moving to such scale, moreover, would also have the welcome side-effect to mitigate minor issues afflicting our indentation measurements, among which adhesion and the non-negligible roughness of the sample. We talked to the dermatologists among our collaborators and ended up with a project that became the capstone of this thesis: a suction-based elastometer for skin, integrated in a handheld probe for 3D depth-resolved elastography. The work is discussed in Chapter 4 and in Appendix 1.

Chapter 1 — Introduction

1.4 A note on Optical Coherence Tomography In this section we will give a brief overview of Optical Coherence Tomography (OCT), the imaging technique that we used in Chapters 3 and 4. We will explain its basic principles, discuss its most relevant specifications, and see how it all applies in the dermatology field. We will also devote a subsection to the combination of OCT and mechanical testing named Optical Coherence Elastography, and we will see why and how that is of special interest. OCT was invented less than three decades ago, yet already found a multitude of applications, especially in ophthalmology (where it is a golden standard), and more in general in the clinical field (Fercher et al. 2003, Swanson & Fujimoto 2017). It is often compared to Ultrasound Echography, with respect to which provides a higher resolution (≤15μm) at the cost of a lower penetration depth (<3mm). The technique generates intensity (greyscale) images, where the source of contrast is the amount of backscattered light from each pixel. A false-color example is in Figure 1.11.

Figure 1.11: False-color B-scan of healthy human skin, with a state-of-the-art Line-Field OCT. “SC: stratum corneum layer; SG: stratum granulosum layer with stretch nuclei; SS: stratum spinosum layer with roundish nuclei; BV: blood vessel; KN: keratinocytes nucleus; DEJ: dermalepidermal junction; H: hair.” from (Levecq et al. 2018)

Those images are the result of the juxtaposition of many 1D axial-scans (Ascans) in one or two transversal directions (B- or C-scans). To obtain the elementary A-scan, light from a broadband coherent source is split into a reference and a sample arm. The light from the sample arm is then focused onto the sample, pointing at increasing depth (perpendicular to the surface). The backscattered light is collected and mixed with the reference arm, giving rise to an interference pattern which is deconvoluted to localize the light intensity coming from each depth, and resulting in the 1-D A-scan. To obtain the two- or three-dimensional tomograms (cross-sections), the optical beam is steered along one, or both, transversal directions.

A note on Optical Coherence Tomography — 1.4

Three OCT specifications are noteworthy for our use case: the resolution, the imaging depth, and the acquisition speed. OCT systems have two independent resolutions: axial and transversal, associated to the different origin of the signal in the two directions. The axial resolution K is related to the bandwidth of the source according to: K =

0.44 K. ΔK

(1.10)

where K and Δλ are respectively the central wavelength of the source and its bandwidth. Note that to keep into account the response of the whole system, Δλ regards the spectrum measured at the detector, not the one at the source. To provide a number, we can apply eq. 1.10 to the system described in Chapter 4, where K = 1310nm, and Δλ = 70nm. Within a sample, the theoretical resolution is further divided by the sample’s index of refraction – for skin Q1 = 1.4(Tearney et al. 1995) – so that we end up with a maximum theoretical resolution of ≈8μm inside skin. In practice, the resolution will always be lower; however, the highest contribution to its decrease – dispersion compensation – can be corrected in postprocessing (Attendu & Ruis 2019). The transversal resolution, diversely, is related to the spot-size and to the precision of the beam steering systems. The penetration depth, i.e. the maximum depth from which the signal can be retrieved, is limited by both intrinsic and extrinsic factors. The characteristics of the light source, the objective and the detector set the upper bounds for theoretical penetration depth. However, when measuring real sample, the penetration depth is primarily limited by attenuation phenomena within the sample (scattering and absorption, the former being practically negligible at 1310 nm), which decrease the amount of backscattered light as depth increases. One of the major complications for OCT in dermatology is skin’s high attenuation coefficient, which severely affects the penetration depth. To mitigate this issue, it is necessary to use light sources at (comparatively) higher wavelengths: our system can reach a 1-2 mm in skin. The acquisition speed is defined as the number of acquired A-scan per second, and depends on the combination of the characteristics of source and detector. However, it is less relevant for our intended application, and our typical speed of 50-100 kHz is more than sufficient. OCT is well suited for dermatology applications where it can provide real-time morphological cross-sections in a non-invasive way and with no sideeffects(Welzel 2001). It has already been used in clinical settings to assess skin structure, and as put by Gambichler et al in a review article, “[OCT] is useful in 41

Chapter 1 — Introduction

visualizing subsurface structures of normal skin, including the epidermis, dermoepidermal junction, dermis, hair follicles, blood vessels and sweat ducts. An increasing number of papers brought evidence of the utility and the precision of OCT technology, in its different technical variants, in diagnosing and monitoring skin disorders, including malignancies and inflammatory conditions, respectively.”(Gambichler et al. 2011); Ulrich et al, goes on to say that “Technical developments of OCT continue to expand the applicability of OCT for different neoplastic and inflammatory skin diseases” (Ulrich et al. 2016).

1.4.1

Optical Coherence Elastography

For all the aforementioned reasons, one can conclude that OCT is a suitable technique for the visualization of dermal tissues, and a meaningful help for dermatologists’ diagnoses. OCT, moreover, is a non-contact technique, and it can be integrated in mechanical probes. Their combination paves the way to a crossdisciplinary technique called Optical Coherence Elastography, which was born with a seminal work from Schmitt(Schmitt 1998) and, as recently reviewed, has been gaining traction ever since (Larin & Sampson 2017). The source of contrast in an OCE image is in the mechanical behavior, as opposed to conventional OCT that captures differences in reflectivity; that contrast enables a completely new imaging modality, one that highlights mechanical differences in the internal structure, independently form reflectivity. Figure 1.12 effectively illustrates this concept in a sample whose features provide mechanical contrast, but no optical one. In a typical OCE experiment, a controlled mechanical stimulus is imparted to a region which is repeatedly imaged by OCT. By comparison of those frames, it is then possible to track and obtain the (spatial distribution of) internal displacements. A mechanical model is then used to estimate the (spatial distribution of) mechanical properties, based on the measured internal displacement, on the known applied load, and on assumptions about the load propagation in the sample. Images that represent the spatial distribution of strain, stress, or a mechanical property, are called Elastograms, an example of which is Figure 1.12(b), where the color intensity represents mechanical properties, thus mapping mechanical contrast, as opposed to conventional OCT that displays the sample’s internal reflectance.

Outline of the thesis â&#x20AC;&#x201D; 1.5

Figure 1.12: a) Conventional OCT, and a) OCE images of the same phantom, that contains inclusions. A transparent polymer on its upper side load allows for the application of the load with simultaneous imaging. The inclusions are made of a different material with respect to the substrate, but have the same reflectivity. As a result, there is no optical contrast for OCT to visualize, while OCE clearly highlights the inclusions. Reproduced from (Kirk et al. 2015)

Elastograms of skin would then be able to highlight all the conditions that affect its mechanical behavior (Liang 2013), including for example scars (Draaijers et al. 2004), burn healing (Brusselaers et al. 2010), cancer (Kirkpatrick et al. 2006), and scleroderma (Enomoto et al. 1996). The synergy of OCT and mechanical testing is particularly appealing; we have directed the work of Chapters 3 and 4 towards the making of devices that could successfully leverage OCE techniques.

1.5 Outline of the thesis So far, we have seen how we approached the diagnostic measurement of the mechanical properties of skin through nanoindentation, but faced severe limitations in reaching that purpose. The techniqueâ&#x20AC;&#x2122;s shortcomings exposed the needs that drove the projects presented hereafter: the outline of this thesis closely follows the points made in the section 1.3.2. Chapter 2 deals with the need for a time-dependent mechanical characterization that, unlike DMA, does not require any preloading, i.e. the nano- R . The chapter 43

Chapter 1 — Introduction

validates a model to compare the two techniques, showing that they correlate as expected. Chapter 3 presents a ferrule-top sensor that integrates an OCT readout. Such multimodal device solves the “blindness” of a nanoindenters to subsurface structures, and allows one to measure mechanical properties and create a depthresolved image with the same sensor and within the same coordinate system. Chapter 4 is about our shift towards measuring the mechanical behavior of skin at the tissue’s millimetric scale, by integrating a suction-based elastometer with a subsurface imaging device. The combination of those techniques paves the way for in-vivo, clinical, optical coherence elastography. In Chapter 5 we discuss a potential business plan to develop the handheld device into a commercial product: we analyze the product-market fit, the landscape of collaborators and competitors, and the financial aspects that would be part of that strategy. In Appendix 1, we present redacted excerpts from the Investigative Medical Device Dossier, needed by the Medical-Ethical commission to allow for clinical research with our handheld probe.

2 Comparison of frequency and strain-rate domain mechanical characterization

Abstract Indentation is becoming increasingly popular to test soft tissues and (bio)materials. Each material exhibits an unknown intrinsic “mechanical behaviour”. However, limited consensus on its “mechanical properties” (i.e. quantitative descriptors of mechanical behaviour) is generally present in the literature due to a number of factors, which include sample preparation, testing method and analysis model chosen. Viscoelastic characterisation – critical in applications subjected to dynamic loading conditions – can be performed in either the time- or frequency-domain. It is thus important to selectively investigate whether the testing domain affects the mechanical results or not. We recently presented an optomechanical indentation tool which enables both strain-rate (nano- R ) and frequency domain (DMA) measurements while keeping the sample under the same physical conditions and eliminating any other variability factor. In this study, a poly(dimethylsiloxane) sample was characterised with our system. The DMA data were inverted to the time-domain through integral transformations and then directly related to nano- R strainrate dependent results, showing that, even though the data do not perfectly overlap, there is an excellent correlation between them. This approach indicates that one can convert an oscillatory measurement into a strain-rate one and still capture the trend of the “mechanical behaviour “of the sample investigated.

This Chapter has been published as: Bartolini, L., Iannuzzi, D., & Mattei, G. (2018). Comparison of frequency and strain-rate domain mechanical characterization. Scientific reports, 8(1), 1-11.

Chapter 2 — Frequency and strain-rate domain mechanical characterization

2.1 Introduction Indentation techniques at the micro- and nano-scale are becoming increasingly relevant in the mechanical characterization of materials, thanks to their numerous advantages such as relatively simple setups, ease of sample preparation, non-invasiveness and high-resolution (M. L. Oyen 2013). While these techniques were initially proposed for metals and ceramic materials (Oliver & Pharr 1992, Bhushan & Koinkar 1994), due to their ability of exerting small and confined loads, they developed into one of the primary tools to investigate soft polymers and biological tissues at typical cell length-scales (D. M. Ebenstein & Pruitt 2006, D. Ebenstein 2010, Farine 2013, Bartolini et al. 2017, Giorgio Mattei, Cacopardo, et al. 2017). It has already been discussed that every material comes with its specific, intrinsic “mechanical behaviour”, which cannot be known a priori (Giorgio Mattei & Ahluwalia 2016). The mechanical behaviour can be measured by means of different testing types and methods. Each measurement provides raw data sets that can be analysed with a given model to obtain specific “mechanical properties”, i.e. quantitative parameters describing the sample mechanical behaviour within the input range investigated (Figure 2.1) (Giorgio Mattei & Ahluwalia 2016). In general, viscoelastic characterization can be implemented either in the time (e.g. creep, stress relaxation, strain rate tests) or frequency domain (e.g. dynamic mechanical analysis), both at the macro- (e.g. bulk testing) and micro-scales (e.g. nano-indentation) (Giorgio Mattei & Ahluwalia 2016). While it is quite straightforward to compare results derived with the same testing method (e.g. DMA tests at different length-scales (van Hoorn et al. 2016)), to relate data obtained in different testing domains (e.g. frequency and time) proves to be more challenging. Zeltmann et al. recently introduced an elegant approach to compare results obtained in the frequency and strain rate domain, and tested the method on data extracted from different sources (Zeltmann et al. 2016). The comparison of data obtained under different conditions, however, is not straightforward. The mechanical results, in fact, may depend on a number of testing and analysis variables, including boundary conditions of the specimen (Jakes & Stone 2011), its preparation (Shen et al. 2005, W. G. Jiang et al. 2008), measurement length-scale (Constantinides et al. 2006, Kaufman et al. 2008, Wrucke et al. 2013) and type of mechanical test (Cohen & Kalfon-Cohen 2013, G. Mattei et al. 2014).

Introduction — 2.1

Figure 2.1: Distinction between sample mechanical behaviour and properties. Acronyms and symbols in figure: DMA = Dynamic Mechanical Analysis; ε ̇M = epsilon dot method; ε = strain; ε ̇ = strain rate; f= frequency; t = time; σ = stress; E= elastic modulus; E'= storage modulus; E''= loss modulus; Eapp= apparent elastic modulus

Even changing the indenter geometry or size has been reported to considerably affect the testing results (Tweedie & Van Vliet 2006, Herbert et al. 2008, Voyiadjis & Peters 2010, Kim et al. 2018). Moreover, the latter are likely dependent on the particular model used to analyse experimental data, because of different assumptions regarding, for instance, material anisotropies, non-linearities (Heymans 2003, M. T. Shaw & MacKnight 2005, F. Fernando 2008), adhesion (Maugis 1992, Buzio et al. 2003), or time-dependency (Strojny & Gerberich 1998, Mencik et al. 2009, Han et al. 2011, G. Mattei et al. 2014). In a recent series of papers, we have presented an optomechanical indentation tool (D. Chavan et al. 2012) that allows one to perform both strain-rate (G. Mattei et al. 2015) and frequency domain (van Hoorn et al. 2016) measurements while keeping the sample under the same physical conditions. To validate the model proposed by Zeltmann and his colleagues, and, therefore, reinforce the bridge between strain-rate and frequency domain measurements, we used our instrument to perform a series of frequency (Dynamic Mechanical Analysis, DMA(van Hoorn et al. 2016)) and strain-rate (nano-epsilon dot method, nano- R (G. Mattei et al. 2015)) measurements on the very same PDMS sample, and, in this way, to selectively study the effect of the measurement domain, while excluding the influence of any other sample-, testing- or analysis-related variable (Giorgio Mattei & Ahluwalia 2016).

Chapter 2 — Frequency and strain-rate domain mechanical characterization

2.2 Methods 2.2.1

Sample preparation

Polydimethylsiloxane (PDMS) samples with flat surfaces were obtained by casting a Sylgard 184 (Dow Corning) pre-polymer solution (50:1 curing agent to elastomer weight ratio, prepared as per manufacturer’s instructions) into a 50 mm diameter Petri dish. The solution was degassed for 15 minutes and cured overnight at 60 °C. The surface of the sample was then passivated with a 5% bovine serum albumin solution (BSA, Sigma Aldrich) in order to minimize adhesion forces (van Hoorn et al. 2016), which invalidate the Hertzian contact assumption of our analyses. Indentation measurements were carried out in phosphate buffered saline (PBS 1x, further reducing tip-sample adhesion (Cappella 2016)) at a controlled temperature of 25 °C. The Poisson’s ratio of PDMS was assumed to be 4 = 0.5 (Mark 2009, G. Mattei et al. 2015, van Hoorn et al. 2016).

2.2.2

Nano-indentation setup

Indentation experiments were performed with a custom setup based on an optomechanical ferrule-top cantilever sensor (S. V. Beekmans & Iannuzzi 2015, van Hoorn et al. 2016), equipped with an R = 261 μm radius spherical tip, measured with a microscope. A commercial interferometer (OP1550, Optics11 B.V.) was used to measure the deflection of the cantilever via Fabri-Perót interferometry(S. V. Beekmans & Iannuzzi 2015). The experimental setup is shown in Figure 2.2. A custom 3D-printed rigid arm was used to mount the probe on a long-range piezoelectric actuator (PI-602.3SL, 300 μm range, Physik Instrumente GmbH), which, in turn, was attached to a manual positioning z-stage manipulator with a 12.5 mm travel length (MVS005, Thorlabs) to coarsely approach the sample before performing the piezocontrolled indentation experiment. The sample was placed on a XY motorized stage actuated by 2 orthogonal linear stages (M664.164 stages, C867 controllers, Physik Instrumente GmbH), which allowed for XY sample positioning and automated surface scanning with an accuracy of less than 1 μm. Indentation experiments were performed at controlled temperature, by enclosing the whole setup in a custom wooden isolation box lined with phonoabsorbent foam, which provides both acoustic and thermal insulation. The temperature inside the box was measured by a PT100 Platinum Resistance Thermometer (Labfacility DM-508, Farnell), connected to a PID controller (Eurotherm 94, ELEKTRO-TRADING) which drove the heating elements (6 × 1 48

Methods — 2.2

Ohm high-power resistors) to maintain a constant temperature during measurements. The stiffness of the ferrule-top cantilever ( ) was determined following a calibration method (assuming a linear spring behaviour) (S. V. Beekmans & Iannuzzi 2015), which resulted in = 8.2 N/m. Given the tip radius (i.e. 248 μm) and the range of sample elastic moduli (~ 25 - 40 kPa), this cantilever stiffness allows to obtain a measurable cantilever bending with good signal-to-noise ratio (thus measurable load) and enough indentation into the sample surface.

Figure 2.2: Nano-indentation setup (adapted from our previous work on multimodal ferruletop sensing (Bartolini et al. 2017)). The inset shows a schematic side view of our ferrule-top cantilever sensor operated by a vertical piezoelectric actuator. The load imparted on the sample surface by the spherical tip is equal to the cantilever stiffness times its deflection (read via interferometry), while the resultant indentation depth into the sample is given by the difference in between piezo-motor displacement downwards and the resultant cantilever deflection upwards.

For instance, testing extremely soft materials with stiff cantilevers would result in a negligible cantilever bending, turning all the piezo movement downwards ( ) into sample indentation (ℎ) and preventing to measure the resultant load via cantilever deflection ( ): please refer to Eq. 2.1 and 2.2 in the following section. Conversely, testing extremely hard materials with very soft cantilever would result in almost equal to , with negligible sample indentation (ℎ). Before each experiment, probe performances were tested by a calibration procedure on 49

Chapter 2 — Frequency and strain-rate domain mechanical characterization

glass in the same conditions used to test samples (i.e. in PBS 1x at 25 °C) (Steven V. Beekmans et al. 2017).

2.2.3

Displacement-, load- and indentation-control mode

Two physical quantities can be measured with the presented setup: i) the load imparted from the spherical tip to the sample surface, and ii) the resultant indentation depth into the sample. According to Hooke’s law, the load ( ) can be derived as the product of the cantilever stiffness ( ) times its deflection ( ) as follows: =

⋅

(2.1)

−

(2.2)

After establishing contact with the sample surface, the indentation depth (ℎ) is given by the difference between the piezo displacement ( ) and the resultant cantilever deflection ( ): ℎ=

The simplest operation mode of our setup is the displacement-control (Dmode), in which we prescribe a given piezo displacement time-profile (S), and measure the resultant (S) and ℎ(S). Notably, D-mode (controlling only the piezo displacement regardless of the cantilever deflection) does not allow a direct control of any of the latter two variables, which is instead needed for dynamic mechanical analysis (van Hoorn et al. 2016) or nano-epsilon dot tests (G. Mattei et al. 2015). To enable load (P-mode) or indentation (I-mode) controlled measurements, we have developed a closed-loop (feedback) control scheme, which reads the cantilever deflection and adjusts the piezo displacement at 500 Hz refresh rate. An on-the-fly determination of the contact point between the indenter tip and the sample surface was implemented to trigger (i.e. switch on) the closed loop control. Notably, the activation of the feedback loop when out-ofcontact results in a divergence of the input signal, jerking the probe into the sample and inevitably breaking the cantilever. In particular, the cantilever deflection was monitored in real-time after starting nano-indentation measurements out of sample contact, and the latter was identified as the instant at which the cantilever bending was higher than 2.5 times the standard deviation of the interferometric readout measured out of sample contact (i.e. instrumental noise). This threshold guaranteed good contact point identification, with negligible onset delay for either the P- or I-mode feedback loop and irrelevant sample pre-stress. Data readout and instrument control were performed with a custom LabView (National Instruments) software. 50

Methods — 2.2

2.2.4

Dynamic mechanical analysis (DMA)

Dynamic mechanical analysis is a technique for material viscoelastic characterization in the frequency domain, which has been applied also to nanoindentation (Herbert et al. 2008, van Hoorn et al. 2016, Steven V. Beekmans et al. 2017). Briefly, DMA is based on the application of a cyclic (sinusoidal) input of displacement or load with frequency T and amplitude ℎ or , respectively, and the measurement of the resultant load or displacement response. If the amplitude of the input oscillation is small enough, the measurements occur within the material Linear Viscoelastic Region (LVR), therefore the stress response is expected to be i) independent of the input amplitude, ii) sinusoidal with the same frequency as the input (T), and iii) present a phase lag =. The latter is 0° for a purely elastic material, 90° for a purely viscous material, and in-between 0° and 90° for any viscoelastic material (Gutierrez-Lemini 2014). In the present study, indentation-controlled (I-mode) DMA experiments were carried out, performing ℎ = 0.30 μm dynamic oscillations around a static ℎ1 = 3 μm indentation depth. These testing parameters satisfy the requirement of small indentation depth with respect to tip radius, offer a good signal-to-noise ratio of the cantilever deflection, eliminate the risk of tip-sample detachment during oscillations, and guarantee the measurement is within the LVR of PDMS (Herbert et al. 2008, van Hoorn et al. 2016). The DMA oscillations were carried out at 5 logarithmically spaced frequencies (T = 0.100, 0.316, 1.00, 3.16, 10.0 Hz). Measurements started out of contact: after an on-the-fly contact identification, the I-mode controlled piezoelectric actuator moved the probe downwards to the static indentation depth ℎ1 = 3 μm in S 2 ,VWX=1s, which was then held constant for S : = 60 s, allowing the sample to reach a new fully relaxed pre-strained equilibrium state (as shown in Figure 2.3C) around which to perform DMA oscillations (starting at S ∗ = S 2 ,VWX + S : = 61s in the schematic shown in Figure 2.3A) (Steven V. Beekmans et al. 2017). Tests at different frequencies were carried out consecutively (with SY = 2 s waiting time in between them, Figure 2.3A), by performing 6 oscillations of the indentation depth and measuring the resultant sinusoidal load response. Experimental indentation and load data obtained at each frequency were fitted to a sinusoid, in order to derive ℎ , and , which were then used to compute the storage, % Z (T), and loss, % ZZ (T), moduli, according to the following equations (Herbert et al. 2008). %Z 1 = cos(=) . (1 − 4 ) ℎ 29ℎ1 &

(2.3)

Chapter 2 — Frequency and strain-rate domain mechanical characterization

%Z′ 1 = sin(=) . (1 − 4 ) ℎ 29ℎ1 &

(2.4)

DMA measurements were performed on n = 36 different (randomly selected) surface points spaced at least by 500 µm, in order to avoid any possible effect from repeated testing cycles on the same spot (G. Mattei et al. 2015, Giorgio Mattei, Cacopardo, et al. 2017). Each measurement comprises all the 5 frequencies of interest, thus resulting in n = 36 replicates per frequency investigated. A custom MATLAB (The Mathworks Inc.) code was used to process the data.

2.2.1

Nano-epsilon dot method (nano- R )

The nano-epsilon dot method (nano- R ) stems from the epsilon dot method ( R ), a technique for material viscoelastic characterization in the strain-rate domain (G. Mattei et al. 2014). The R is based on performing mechanical tests at different constant strain-rates, reporting a strain-rate dependent apparent elastic modulus, % ( R). The latter is defined as the slope of the stress-strain response of the material within the LVR and increases with strain-rate ( R) in case of viscoelastic materials.

The R has been recently adapted from bulk testing to nano-indentation, to obtain the nano- R , which introduces a new definition of indentation stress (") and strain ( ) based on the Hertzian contact (G. Mattei et al. 2015). "=

&√ℎ&

(2.5)

4 ℎ \ ] . 3(1 − 4 ) &

(2.6)

^ 4 ℎR = ` a . ^S 3(1 − _ ) &

(2.7)

Thanks to the new definitions (Eq. 2.5 and 2.6), the indentation strain-rate ( R = ^ /^S) is independent of the indentation strain ( ) and linearly proportional to the indentation rate (ℎR) as follows: R=

Therefore, to implement a constant strain-rate test at a desired R b , one has simply to impose a constant indentation rate (ℎR), obtained by reverting Eq. 2.7. 52

Methods — 2.2

In this work, tests were performed at 5 logarithmically spaced strain-rates chosen as representative of the 5 logarithmically spaced frequencies used for DMA experiments (0.100 - 10.0 Hz). Notably, the conversion of those frequencies to a constant strain-rate is inherently impossible, because the strain-rate is continuously varying during oscillations at a constant frequency. Indeed, the sinusoidal oscillation of the indentation input results in a co-sinusoidally varying strain-rate.

Figure 2.3: Prescribed indentation profiles for A) DMA and B) nano-ε ̇M measurements. C) Experimental data showing that the load plateau after ramping to hs = 3 μm was reached in less than 20s.

Chapter 2 — Frequency and strain-rate domain mechanical characterization

To choose 5 representative constant strain-rates for the nano- R , we calculated the instantaneous strain-rate in the quasi-linear portion at the beginning of the oscillation, i.e. the first time derivative of the oscillating strain across the static indentation depth ℎ1 = 3 μm around which oscillations with amplitude ℎ = 0.30 μm and frequency T are performed. Considering the indentation input equal to ℎ = ℎ1 + ℎ cdQ(2eTS), the strain rate (defined as in Eq. 2.7) is given by R=

4 ^ℎ 4 = 2eTℎ fgc(2eTS) . 3&(1 − _ ) ^S 3&(1 − _ . )

(2.8)

Assuming fgc(2eTS) = 1 in the quasi-linear portion around the static oscillation depth ℎ1 , the former equation 2.8 simplifies to: R=

4 T⋅ℎ 2eTℎ ≈ 11.18 ⋅ . 3&(1 − _ ) &

(2.9)

from which representative strain-rates for nano-epsilon dot measurements were derived (by substituting ℎ = 0.30 μm; & = 248 jk; f = 0.100, 0.316, 1.00, 3.16, 10.0 Hz), obtaining R = 0.00135, 0.00427, 0.0135, 0.0427, 0.135 s-1.

Nano- R measurements started out of contact. After finding the contact point on-the-fly, the I-mode controlled piezoelectric actuator moved the probe downwards at the desired indentation rate ℎR up to ℎ1 = 3 μm depth (i.e. the static indentation depth value used in DMA experiments; S 2 in Figure 2.3B). Consistently with the DMA testing protocol, this indentation depth was held constant for S : = 60 s (Figure 2.3C), to allow the sample to reach the same, fully relaxed, pre-strained, equilibrium state as in DMA measurements. The relaxed and pre-strained material was then probed with a second constant strainrate test at the same ℎR (thus R) as the first ramp, to obtain pre-strained apparent elastic moduli, that are meaningfully comparable with those derived from DMA measurements.

Experimental load-indentation ( − ℎ) data were converted into stress and strain according to Eq. 2.5 and 2.6, respectively. Data belonging to the first ramp were used to derive the strain-rate dependent apparent elastic moduli for the virgin material (% ,m ( R), obtained in absence of pre-strain) as the slope of the first linear tract of the " − curve (G. Mattei et al. 2015, Giorgio Mattei, Cacopardo, et al. 2017), which identifies the absolute linear viscoelastic region (LVR) (G. Mattei et al. 2014). Strain-rate dependent elastic moduli for pre-strained material, % ,n ( R), were derived in the same manner from data belonging to the second ramp, i.e. within a local LVR. Notably, the measurement on the prestrained material started at the same level of pre-strain necessary for DMA 54

Methods — 2.2

experiments (ℎ1 = 3), and in the same equilibrium relaxed state. The % ,n ( R) data can thus be meaningfully compared with DMA results (G. Mattei et al. 2014, Giorgio Mattei & Ahluwalia 2016).

Nano- R measurements were performed on n = 30 different (randomly selected) surface points spaced at least by 500 µm, obtaining n = 6 replicates for each of the 5 strain-rates investigated. Data were processed with a custom MATLAB (The Mathworks Inc.) code.

2.2.2

Comparing DMA and nano- R

results

To compare results obtained from frequency-domain (DMA) and strain-rate domain (nano- R ) experiments, a storage modulus master curve was extrapolated by fitting experimental % Z (T) data from DMA to the following sigmoidal function of ln(p) (Eq. 2.10) (Zeltmann et al. 2016): % Z (p) = F ∙ tanh6u ∙ (ln(p) + f)7 +

(2.10)

where p = 2eT represent the angular frequency, while F, u, f and represent the fit coefficients. This sigmoidal fitting function assumes that the mechanics of the tested material can be described by a simple rheological behaviour with one characteristic relaxation frequency (or time) - a good assumption for PDMS (D. Jiang et al. 2012, A. Tirella et al. 2014, G. Mattei et al. 2015, Niu et al. 2016). In particular, the storage modulus master curve presents only one smooth step transition, corresponding to one peak in the loss modulus frequency spectrum, and the behaviour is asymptotic when going to either zero or infinity frequency. This function satisfies the physical requirement of positive and bounded behaviour of the relaxation function at both zero and infinity frequency if > F (Christensen & Freund 1984). In case of multiple % Z smooth transitions (and concomitant % Z ′ peaks), a combination of sigmoidal functions like Eq. 2.10 is needed to describe the experimental behaviour and yield the entire relaxation function (Zeltmann et al. 2016, 2017), while keeping the analysis steps below unaltered. The frequency-domain storage modulus function obtained from the fitting, % Z (p), was then converted into its respective time-domain relaxation modulus function, %(S), by solving numerically the following integral from the linear theory of viscoelasticity (Christensen & Freund 1984, Zeltmann et al. 2016, 2017) %(S) =

2 x % Z (p) cdQ(pS) ^p w e p

(2.11)

Chapter 2 — Frequency and strain-rate domain mechanical characterization

The stress-time response to a given strain history can be derived from the timedomain relaxation function according to the following convolution integral (Christensen & Freund 1984): "(S) = % ∗ ^ = w %(S − y)

where " denotes the stress, integration.

^ (y) ^y ^y

(2.12)

the strain, and y a time variable used for

For a constant strain-rate deformation input convolution integral in Eq. 2.12 simplifies to: "(S) = R w %(y) ^y

R beginning at S = 0, the (2.13)

The stress-strain response was then obtained by numerical integration of "(S) from Eq. 2.13, followed by a linear transformation of the independent variable time (S) into strain ( = S ∙ )R .

Stress-strain responses were computed for each of the 5 strain-rates investigated with the nano- R and then used to calculate DMA-derived strainrate dependent apparent elastic moduli, % ,VWX ( R), as stress-strain slope within the same LVR used for % ,n ( R) values from nano- R measurements. Thanks to this procedure, % ,VWX ( R) and % ,n ( R) data can be directly and meaningfully compared. Calculations to obtain strain-rate dependent stress-strain responses from DMA frequency-dependent data were performed in Mathematica (Wolfram Research Inc., IL, USA).

2.3 Results 2.3.1

Dynamic Mechanical Analysis (DMA)

Storage (% Z ) and loss (% ZZ ) moduli (Figure 2.4A) were measured at 5 different logarithmically spaced frequencies (T = 0.100, 0.316, 1.00, 3.16, 10.0 Hz), performing ℎ = 0.3 μm amplitude oscillations around a static ℎ1 = 3 μm indentation depth(van Hoorn et al. 2016) (see Methods section for details).

A storage modulus master curve was derived by fitting experimental % Z (T) data to a sigmoidal function (Eq. 2.10). Notably, this function is not intended to represent a specific viscoelastic model, but rather it was used here only as a mean to extrapolate a physically-relevant storage-modulus master-curve from

Results — 2.3

experimental data. That master-curve was then used to compute the time-domain relaxation function, necessary to compare frequency (DMA) and strain-rate (nano- R ) domain results as described in the Methods section. Fitting results were: F = 196.9; u = 0.2006; f = -11.74; = 220.4 (Figure 2.4B).

2.3.1

Nano-epsilon dot (nano- R )

Nano-epsilon dot measurements were performed at 5 different logarithmically spaced strain-rates ( R = 0.00135, 0.00427, 0.0135, 0.0427, 0.135 s-1). The nanoR has been originally developed to derive virgin material properties, i.e. in absence of any pre-strain/stress. However, nano- R measurements were started at the same level of static pre-strain required by DMA (i.e. ℎ1 = 3 μm), to obtain meaningfully comparable results between the two methods. Measurements were also performed in absence of pre-strain, to relate virgin and pre-strained apparent elastic moduli: % ,m ( R) and % ,n ( R), respectively. The linear viscoelastic region (LVR) extended up to 0.0175 strain in both sample conditions with strainrate-increasing moduli (as expected for viscoelastic materials) exhibiting similar trends (Figure 2.5).

Figure 2.4: Dynamic mechanical analysis results obtained for PDMS. A) Experimental storage (% Z , blue circles) and loss (% ZZ , orange triangles) moduli versus frequency (dotted lines represent linear data interpolation and only serve as a guide to the eye). B) Storage modulus master curve (dashed line) obtained by fitting experimental % Z (T) (blue circles) to the sigmoidal function % Z (p) = F ∙ tanh6u ∙ (ln(p) + f)7 + , needed to derive time-domain results from DMA. Error bars denote standard errors.

Chapter 2 — Frequency and strain-rate domain mechanical characterization

2.3.2

Comparing frequency and strain-rate domain results

The storage modulus master curve obtained fitting experimental % Z (T) data from DMA was integrated numerically according to Eq. 2.11 to derive the time-domain relaxation-modulus function, %(S), needed to compare frequency-domain (DMA) and strain-rate domain (nano- R ) results. Since %(S) ≥ 0, ^%(S)/^S ≤ 0 and ^ . %(S)/^S . ≤ 0, the computed %(S) solution satisfies the requirements of fading memory and non-negative stored and dissipated energy in the whole time range 0 to ∞ (Christensen 1972, Christensen & Freund 1984). The %(S) was then integrated according to Eq. 2.13 to obtain the stress-time responses to the constant strain-rate inputs of interest (i.e. the same used for nano- R measurements, namely R = 0.00135, 0.00427, 0.0135, 0.0427, 0.135 s-1). The integral was solved numerically up to S =13 s with 2 ms time steps, to have enough data and resolution to derive the stress-strain response within the LVR strain range ( = 0 - 0.0175). Indeed, computing the stress-strain response within LVR for the lowest strain-rate of interest requires solving up to a time S|m} = R = 0.0175/0.00135 ≈ 12.96 s , while for the highest strain-rate 2 :/ 2 solving up to ~ 0.13 s is required. Eventually, the DMA-derived apparent elastic moduli, % ,VWX ( R), were computed as the stress-strain slope within the same LVR used for nano- R , and compared to the % ,n ( R) resulting from pre-strained nano- R (Figure 2.5). This framework allowed us to meaningfully compare strain-rate and frequency derived mechanical results, and to investigate the specific effect of the testing domain

Results â&#x20AC;&#x201D; 2.3

Figure 2.5: Strain-rate dependent apparent elastic moduli obtained for virgin (% ,m blue circles) and pre-strained (% ,n orange triangles) PDMS. Dashed lines are linear interpolation and only serve as a guide to the eye. Error bars denote standard errors.

Figure 2.6 shows an increase of % ,VWX with strain-rate, as expected for viscoelastic materials and in agreement with experimental % ,n ( R) results. Overall, the % ,n ( R) values were higher than their respective % ,VWX ( R) ones, regardless of the strain-rate. However, an optimal correlation was observed between % ,VWX ( R) and % ,n ( R) data (r = 0.99), with an almost constant difference in the apparent moduli, regardless of the strain-rate.

Chapter 2 â&#x20AC;&#x201D; Frequency and strain-rate domain mechanical characterization

Figure 2.6: Strain-rate dependent apparent elastic moduli at the same level of pre-strain (i.e. 3 Îźm indentation depth), as obtained by nano- R (% ,n , orange triangles) and derived from DMA (% ,VWX , blue circles). Dashed lines are linear interpolation and only serve as a guide to the eye. Error bars denote standard errors. The inset shows the correlation plot between % ,VWX ( R) and % ,n ( R) data (r = 0.99).

2.4 Discussion The investigation of the testing domain effect on the ensuing mechanical results is of primary interest in material engineering and design (Zeltmann et al. 2017). Despite the large amount of DMA data available in the literature for several materials, frequency-domain results obtained with this method have rarely been used to design structures and components, mainly because they are not directly applicable to most of engineering problems. Conversely, mechanical properties derived in the strain-rate domain will be more convenient in such applications, but they are generally more challenging to obtain mostly because of limited speed ranges attainable with a given testing setup, and the quite long experimental trials needed to perform measurements at very low strain-rates, which result in time consuming and limited throughput tests (Zeltmann et al. 2016).

Frequency-dependent storage (% Z ) and loss (% ZZ ) moduli were obtained from DMA measurements at 5 different log-spaced frequencies (T = 0.100, 0.316, 1.00, 3.16, 10.0 Hz) on PDMS samples. In particular, both % Z and % ZZ increased with frequency, consistently with previous DMA results obtained at micro- and macroscale (Hisyam A. Razak et al. 2015, van Hoorn et al. 2016). In the assumption of simple rheological behaviour with one characteristic relaxation frequency (or 60

Discussion — 2.4

time) for PDMS (D. Jiang et al. 2012, A. Tirella et al. 2014, G. Mattei et al. 2015, Niu et al. 2016), one could expect a peak in the % ZZ spectrum in correspondence of that frequency and a correspondent smooth transition in % Z spectrum, from the asymptotes of i) equilibrium (or relaxed) modulus at low frequencies, to ii) instantaneous modulus at high frequencies (G. Mattei et al. 2014). Since % Z and % ZZ show no asymptotic plateau and % ZZ keeps increasing with frequency, the material relaxation frequency is expected to be outside and to the right of the 0.1 - 10 Hz range investigated, in agreement with previous works (Hisyam A. Razak et al. 2015, van Hoorn et al. 2016).

Nano- R measurements were performed at 5 different log-spaced strain-rates ( R = 0.00135, 0.00427, 0.0135, 0.0427, 0.135 s-1). Notably, nano- R measurements started at the same level of static pre-strain needed by DMA to perform dynamic oscillations (i.e. 3 μm indentation depth in this study), in order to measure the sample in the same fully-relaxed pre-strained equilibrium state and rule-out any depth-related nonlinearity (Kaufman & Klapperich 2009, Charitidis 2011, G. Mattei et al. 2014, Giorgio Mattei & Ahluwalia 2016). Since the nano- R does not require any static pre-strain to start measurements (G. Mattei et al. 2015), it was also employed to derive virgin (i.e. not pre-strained) properties. Both the virgin (% ,m ) and pre-strained (% ,n ) apparent elastic moduli exhibited a similar increasing trend with strain-rate ( R), as expected for PDMS at micro- and macro-scales (Khanafer et al. 2009, A. Tirella et al. 2014, G. Mattei et al. 2015). A simple rheological behaviour with one characteristic relaxation time (as assumed for DMA) results in an apparent elastic modulus increasing with strain-rate, from an asymptotic equilibrium value at low strainrates to a higher instantaneous one (Lakes 2009, G. Mattei & Ahluwalia 2019). While a slow increase in both % ,m and % ,n was observed at low strain-rates, suggesting that these moduli are approaching the respective equilibrium values expected for R → 0, no upper plateau values were observed with increasing strain-rates, suggesting that the transition to the instantaneous modulus would be completed beyond the highest strain-rate investigated (0.135 s-1 ), in agreement with other reports (A. Tirella et al. 2014, G. Mattei et al. 2015). Notably, the % ,m ( R) values were higher than % ,n ( R) ones, regardless of the strain-rate. This result is consistent with Charitidis, reporting on the existence of a PDMS surface/near surface region (extending up to ∼2.5 μm indentation depth) characterized by higher elastic moduli that significantly decrease with penetration depth and plateau to a lower bulk value (Charitidis 2011). The closed-loop (feedback) control scheme enables constant indentation rate (ℎR) measurements, which represents a notable advancement with respect to the open loop control (D-mode) employed in previous nano- R tests(G. Mattei et al. 2015). Indeed, although in a D-mode controlled measurement a constant piezo 61

Chapter 2 — Frequency and strain-rate domain mechanical characterization

displacement rate results in a nearly constant strain-rate ( R) within the region of small deformations, the actual R experienced by the sample is generally decreasing over time, and lower than the value calculated by imposing ℎR equal to the piezo displacement rate in Eq. 2.7 (G. Mattei et al. 2015, Giorgio Mattei, Cacopardo, et al. 2017). In particular, in a constant piezo displacement rate experiment, the higher the indentation depth into the sample surface, the higher the cantilever deflection rate because of the increased contact area between its spherical tip and the sample, which, in turns, results in a lower actual indentation rate (and strain-rate) experienced by the sample as the tip advances during measurements (G. Mattei et al. 2015, Giorgio Mattei, Cacopardo, et al. 2017). The closed-loop control scheme solves this problem, by adapting the piezo displacement rate during measurements in order to enable constant indentation rate experiments (I-mode), resulting in a constant R experienced by the sample over the entire indentation range. This is also critical to perform nano- R measurements at a given level of pre-strain, required, for instance, to investigate the strain-dependency of mechanical properties and to compare nano- R and DMA results, as presented in this work. We obtained apparent elastic moduli at different strain-rates in two ways: directly from nano- R measurements, % ,n ( R), and indirectly after integration of DMA data, % ,VWX ( R). We found an excellent correlation between them (r=0.99, insert Figure 2.6), with a constant undershoot of % ,VWX ( R) with respect to % ,n ( R). The average deviation between % ,n and % ,VWX ,equal to -10.26 ± 1.02 % (mean ± std. err.), is consistent with that reported by Zeltmann et al. (Zeltmann et al. 2016, 2017), and indicates an overall good agreement between frequency and strain-rate domain results. Such discrepancy indicates the presence of a systematic error, possibly owed to the narrow frequency range attainable by our testing setup. Notably, outside of that 0.1 - 10 Hz range, the accuracy of the time-domain relaxation function depends on the extrapolation of the fitted storage-modulus master function. Possible % Z transitions (and concomitant % ZZ peaks) outside the measured frequency range are indeed not considered and may cause deviations from the true relaxation function of the material under testing. Given the maximum frequency of 10 Hz investigated in this work, the smallest time from which the relaxation function reflects experimental data can be estimated as S → 1/T (Christensen & Freund 1984), corresponding to a lower limit of 0.1 s; at shorter times, the relaxation function is primarily influenced by the extrapolation of the storage modulus fit at frequencies > 10Hz. At the maximum strain-rate investigated in this work, the upper limit of the LVR is reached in S ƒ = 2 : / 2 R : = 0.0175/0.135 ≈ 0.13 s, meaning that the respective % ,VWX value is largely derived from extrapolated data (though all times are affect to some extent by all frequencies, and vice versa). 62

Conclusion â&#x20AC;&#x201D; 2.5

However, since the difference between % ,n and % ,VWX values was almost independent of the strain-rate, there is good evidence for a systematic error in the evaluation of DMA moduli. We investigated this possibility by a variation analysis on the parameters of the fit. In particular, a 2% increase in the parameter d (i.e. the offset addend in Eq. 2.10) results in an almost perfect match between % ,n and % ,VWX , and a small upward transition of the master curve. If systematic error on DMA had the same magnitude, there would be no discrepancy between % ,n and % ,VWX .

2.5 Conclusion In this work we show a direct comparison between mechanical results derived from strain-rate and frequency domain measurements, in a setting where â&#x20AC;&#x201C; for the first time â&#x20AC;&#x201C; any other sample-, testing- and/or analysis-related source of variability was eliminated (Giorgio Mattei & Ahluwalia 2016). The demonstration that testing in either the frequency or strain-rate domain returns compatible results is critical towards a more comprehensive understanding of material viscoelastic (i.e. time-dependent) behaviour, which is of primary importance for a number of applications, ranging from mechanical and structural design for civil and (bio)material engineering (Jelen et al. 2013, Leta et al. 2015), to tissue engineering (Annalisa Tirella et al. 2015, Giorgio Mattei, Magliaro, et al. 2017) and cell mechano-biology (Mammoto et al. 2013, Giorgio Mattei et al. 2015).

3 Multimodal probe for Optical Coherence Tomography epidetection and micron-scale indentation

Abstract: We present a multimodal ferrule-top sensor designed to perform integrated epidetection of Optical Coherence Tomography depth-profiles and micron-scale indentation, by all-optical detection. By scanning a sample under the probe, we can obtain structural cross-section images and identify a region-of-interest in a non-homogeneous sample. Then, with the same probe and setup, we can immediately target that area with a series of spherical-indentation measurements, in which the applied load is known with a μN precision, the indentation depth with sub-μm precision and a maximum contact radius of 100 μm. Thanks to the visualization of the internal structure of the sample, we can gain a better insight into the observed mechanical behavior. The ability to impart a small, confined load, and perform OCT A-scans at the same time, could lead to an alternative, high transverse resolution, OCE sensor.

This Chapter has been published as: Bartolini, L., Feroldi, F., Weda, J. J. A., Slaman, M., De Boer, J. F., & Iannuzzi, D. (2017). Multimodal probe for optical coherence tomography epidetection and micron-scale indentation. Journal of Innovative Optical Health Sciences, 10(06), 1742007

Chapter 3 â&#x20AC;&#x201D; Multimodal probe for OCT and indentation

3.1 Introduction: The importance of mechanical measurements is undisputedly recognized in several research areas. In the field of biomedical research, for instance, quantitative assessments of the viscoelastic properties of cells and tissues allow scientists to investigate the role of mechanics in the physiology of living systems(Holzapfel & Ogden 2006, Cowin & Doty 2007, Wells 2013). This kind of studies may eventually lead to major breakthroughs in tissue engineering and in the detection of life-threatening diseases(Lekka et al. 2012, Lokody 2014). The local mechanical properties of biological materials are typically assessed via Atomic Force Microscopy (AFM)(Binnig et al. 1986, Crichton et al. 2013, M. L. Oyen 2013, Pittenger & Scientist 2013), by means of so-called nanoindentation techniques. A tip, mounted at the end of a micromachined cantilever spring, is pushed into the sample with a calibrated stroke. By looking at how the deflection of the cantilever evolves over time, one can infer the mechanical properties of the material underneath the indented point(M. L. Oyen 2006, G. Mattei et al. 2015). Indentation techniques, however, suffer from a main limitation: they measure the collective behavior of the volume of material underneath the contact area: when researchers observe spatial variations of mechanical properties, they cannot tell whether those differences arise from the presence of heterogeneous structures in the sensed volume, or if that volume is homogeneous and simply possesses different mechanical properties. This ambiguity is particularly detrimental when indentation is used as a diagnostic tool to discern healthy tissues from diseased ones. Biological samples, indeed, typically present a high degree of multiscale structures which affect their macroscopic behavior(Katz et al. 2007, Guo et al. 2013, Egan et al. 2015). A visualization of the inner structures of the sample would allow for a better interpretation of mechanical measurements and for a disambiguation between the influence of internal structures and intrinsic mechanical differences. A noteworthy technique for this purpose is Optical Coherence Tomography (OCT), which offers a field-of-view and a resolution relevant to the scales of nano- and micro-indentation. In 1998, Schmitt published the seminal paper of Optical Coherence Elastography (OCE)(Schmitt 1998), in which he combined OCT imaging to assess strain in the material and mechanical loading to impart a known compressive stress, and eventually obtain a visualization of the mechanical properties of a sample, called elastogram. There are now multiple approaches to the creation of elastograms by OCE, differing in the type and spatial extent of the loading mechanism, or in its 66

Introduction: — 3.1

temporal characteristics(Brendan Francis Kennedy et al. 2014, Shang Wang & Larin 2015, Mulligan et al. 2016, Brendan F. Kennedy et al. 2017, Larin & Sampson 2017). To aim at applications in life sciences researchers put considerable effort in increasing the resolution of elastograms, targeting the finer structures of tissues that may be early indicators of disease. Recent OCE works obtained microscale resolution(Brendan F. Kennedy et al. 2014, Chin et al. 2014, K. M. Kennedy et al. 2015, Es’haghian et al. 2017) elastograms, improving the OCT signal processing to extract more accurate strain measurements and reducing the contact area to apply a more localized mechanical stimulus; effort in miniaturization resulted in needle-based OCE for in-vivo and in-situ characterization(K. M. Kennedy et al. 2012, 2013, Qiu et al. 2016). Over the last few years, our group has pioneered a new indentation technique that has been tailored for applications in life science research(D. Chavan et al. 2012, S. V. Beekmans & Iannuzzi 2016, van Hoorn et al. 2016). The technique relies on a ferrule-top probe, which is obtained by assembling a millimeter-size cantilever, equipped with a ≈100 μm diameter sphere on its free hanging end, on a small glass block. The block hosts an optical fiber used to measure the deflection of the cantilever and reproduce the indentation protocol used in AFM nanoindentation. This approach has already been used by several groups to assess the mechanical properties of tissues and cells (Neufurth et al. 2014, Shunfeng Wang et al. 2014, G. Mattei et al. 2015, Müller et al. 2015, Moshtagh et al. 2016, Bos et al. 2017, Lavet & Ammann 2017, Sarker et al. 2017, Vashaghian et al. 2017). Furthermore, in 2013, we have demonstrated that the ferrule-top probes used for indentation can be modified to host another optical fiber that, connected with an OCT system, enables the user to look at how the subsurface features of a sample deform when the indenter is pushed into the sample(Dhwajal Chavan et al. 2013). Those studies have never been brought beyond a first proofof-concept, whose practical relevance in life sciences was very limited: the use of new fabrication techniques and materials lead to a reduction of the stiffness of the cantilever by 3 orders of magnitude with respect to those initial studies, enabling measurements on biologically-relevant soft materials. Moreover, the old design was imparting the compressive load by means of a small tube glued to the cantilever, which allowed the OCT signal to pass through. Unfortunately, such indenter geometry posed severe limitations to the quantitative analysis of the mechanical data, its sharp edges lead to stress accumulation and could have damaged soft samples and there is no analytical model for such an indenter shape. This issue was crucially overcome as the new, transparent cantilever allows us to substitute the tube with a hemispherical sapphire indenter tip, which at the same time, acts as a half-ball lens, for focusing and epidetection of 1D OCT depth profiles (A-scans) along the indentation axis.

Chapter 3 — Multimodal probe for OCT and indentation

Our probe can hover over the surface of a thick sample mounted on motorized stages: by driving them in a raster scan under the fixed sensor we can obtain cross-sections or volumetric OCT images (B- or C-scans), with which visualize and identify possible subsurface structures. The probe can then target regions-ofinterest and perform quantitative microindentation measurements in there. Other designs combining OCT and microindentation have been published, but they are only applicable to thin samples, since indentation and OCT are performed from different sides of sample(Yang et al. 2007). The integration of indentation and OCT on the same sensor allows for an immediate and accurate co-localization of the optical and mechanical measurements, with notable gains in speed and repeatability. Moreover, the geometry of the indenter makes it possible to test virgin materials (no need for precompression) at very fine scales, keeping the contact radius below 100μm. We show that, with this system, we can indeed acquire B-scans of a sample that hides complex features underneath a homogenous surface, and that the probe is then capable of distinguishing different mechanical responses when positioned on top of different subsurface structures.

3.2 Experimental details: 3.2.1

Ferrule-top sensor: design and fabrication

Figure 3.1 shows a microscope image of the probe, which was designed to combine ferrule-top indentation with the acquisition of OCT depth-profiles in an epidetection scheme. The probe consists of a borosilicate glass ferrule hosting: (i) a cantilever spring, used to apply a calibrated mechanical load, (ii) a half-ball lens, used both as indenting tip and as focusing element for the OCT signal, and (iii) two optical fibers, one used to measure the deflection of the cantilever and the other used to perform OCT. The fabrication of the probe consists of three steps: preparation of the ferrule, preparation of the cantilever with the lens, and their assembly. In the first step, a 3 mm x 3 mm x 7 mm borosilicate glass ferrule is mounted on a diamond wire cutter, which is used to carve a 3 mm x 500 μm x 500 μm ridge out of the small facet of the ferrule and a deep groove on its side (Figure 3.2A).

Experimental details: — 3.2

Figure 3.1: Microscope side view of the ferrule-top sensor combining indentation and OCT capabilities.

In the second step (Figure 3.2B), a borosilicate glass ribbon (Vitrocom Inc), with a rectangular section of 30 μm x 300 μm, is made reflective for most of its length by sputtering a 5 nm thick chromium adhesion layer followed by a 100 nm thick gold film. The remaining length of the ribbon, around 500 μm, is left transparent. There, a sapphire half-ball lens (Edmund Optics, 200 μm radius) is glued on the opposite face of the coating. In the third step (Figure 3.2C), the ribbon is glued to the ridge of the ferrule. A single mode optical fiber (Corning SMF28) is then cleaved and mounted in the side groove, pointing at the reflective part of the cantilever. Another single mode optical fiber (Corning SMF28), cleaved at an angle of 8° to suppress back reflections, is glued on the long face of the ferrule, aligned with the lens at the free hanging end of the cantilever.

Chapter 3 — Multimodal probe for OCT and indentation

Figure 3.2: OCE Ferrule-top fabrication steps. The ferrule is showed upside-down to ease the visualization. A) preparation of the ferrule (micromachining of the ridge and of the side groove); B) preparation of the cantilever; and C) assembly of the two readout fibers and of the cantilever.

3.2.2

Setup

The first optical fiber, mounted in the side groove and aligned with the reflective part of the cantilever is connected to a commercial interferometer (OP1550, Optics11) to measure, via Fabry-Pérot interferometry, the deflection of the cantilever. The other optical fiber is connected via a circulator (CIR-1310-50-APC, Thorlabs Inc., Newton, NJ, USA) to the sample arm of a custom swept-source based OCT system that has been described in previous reports (J. Li et al. 2012, 2015). Light from a 1310 nm swept source laser (A-line rate = 50 kHz, optical bandwidth = 90 nm, Axsun Inc., Billerica, MA, USA) is split into a sample arm and a reference arm, set in a Mach-Zehnder modality. About 12 mW of sample arm light reach the sensor through the second port of a circulator, which in turn receives the backscattered light collected by the ball lens. The light from the circulator’s third port interferes with the reference arm light and the so-called DC terms are rejected through balanced detection (PDB430C, Thorlabs Inc., Newton, NJ, USA), yielding a sensitivity of 112 dB. The acquisition of a B-scan (2,000 A-lines/B-scan) starts when the position of the stage reaches a set location: the stage controller then sends an acquisition-trigger to the high-speed digitizer (ATS9350, Alazar Inc., Pointe Claire, Canada). The data are processed as follows; the average k-space spectrum is subtracted from each individual spectrum to remove DC noise and fixed patterns in the image. Then, a cosine window and a chromatic dispersion correction curve are applied before performing a Fourier transform (fft, Mathworks Inc., Natick, MA,

Experimental details: — 3.2

USA). Finally, the resulting absolute values are converted in dB scale, producing an A-line. The ferrule-top probe is mounted on a 3D printed rigid arm, as shown in Figure 3.3. The arm is fixed to a long-travel piezoelectric device (PI-602.3SL, 300 μm range, Physik Instrumente GmbH), which in turn is fixed on a coarse z-axis manipulator with a travel of 12.5 mm. The sample is placed on top of two orthogonal linear stages (M664.164, Physik Instrumente GmbH, driven by C867 controllers from the same company) that allow for X-Y positioning with an accuracy of less than 1 μm.

3.2.1

Experimental procedure

To acquire an OCT image, the tip of the sensor is initially brought approximately 800 μm above the surface of the sample. Using the XY stages, the sample is then moved back and forth to allow for B-scans and, if required, C-scans. The maximum speed for the fast-scan axis is 75 mm/s. In order to guarantee a regime of constant speed, the motion includes a run-up of 2 mm, after which a hardware trigger from the XY stage controller initiates the acquisition of the B-scan. Maximum repeatability of the B-scans is guaranteed by unidirectional scanning. The axial resolution of the OCT signal depends on its light source, and is around 10 μm in PDMS, whereas the lateral resolution (roughly given by the spot size, determined by the geometry of the optical components of the probe) is estimated by Zemax OpticStudio simulations to be around 30 μm. Before the indentation experiment, we bring the sensor close to the surface with the Z-axis manipulator and extend the piezoelectric device until the tip gets in contact with the surface. Then the actual mechanical measurement begins: as the tip is pushed into the sample, we record the extension of the piezoelectric translator (via its internal strain-gauge), the deflection of the cantilever (via the interferometer), and the indentation depth (calculated as difference between the piezoelectric extension and the cantilever deflection). Using a software feedback loop running at 500Hz, we impose a temporal profile to the indentation depth by adjusting the position of the piezoelectric element. Eventually, we calculate the force required to reach a predefined depth set-point by multiplying the measured deflection of the cantilever by the associated spring constant, previously obtained via a standard calibration technique(S. V. Beekmans & Iannuzzi 2015).

Chapter 3 — Multimodal probe for OCT and indentation

Figure 3.3: Overview of the setup. OCT imaging is obtained by moving the sample with the XY stages while the sensor hover at approximately 800 μm above the sample. Indentation measurements make use of the interferometric readout and are driven by the piezoelectric element.

3.2.2

Sample preparation

To demonstrate the working principle of this ferrule-top sensor, we created a sample that, while avoiding the ethical and technical issues of real biological tissues, shares with them several meaningful properties. Such tissue-mimicking phantom was made out of a bi-component crosslinked elastomer polydimethylsiloxane (PDMS, Sylgard 184, Dow Corning) and presents the following characteristics: • • • •

Layers: the sample is composed of two layers Structure: a step defect changes the thickness of the top layer Mechanical contrast: the top layer is stiffer than the bottom one Optical contrast: the bottom layer scatters more than the top one

The preparation of the sample borrows from a technique used in soft lithography(Walia et al. 2015), in which a positive mask defines the structure imprinted on the bottom surface of a casted polymer. We start by gluing a coverslip (thickness of 100μm) to a microscope slide and spray it with a thin layer of Teflon (DuPont, dry-film lubricant), in order to prevent adhesion later on. We place the microscope slide at the bottom of a mold and cast the prepolymer of PDMS inside it. After curing, we peel the polymer off the mask and out of the mold and put it upside-down in the mold, where it is covered with another PDMS layer. 72

Results and discussion: â&#x20AC;&#x201D; 3.3

Following this procedure, we are able to obtain a sample with a flat surface and a non-homogeneous subsurface structure. To create mechanical contrast, the bottom layer is made with a ratio of 40:1 elastomer to curing agent, whereas the top layer is tuned to be stiffer by mixing the components in a ratio of 10:1 (higher fraction of curing agent results in more crosslinks and higher stiffness). Furthermore, by substituting different fractions of the elastomer with a special solution (made with TiO2 particles suspended in OH-terminated silicone oil), we can obtain different scattering coefficients in the two layers, resulting in optical contrast for the OCT image. As expected, the finished sample appears homogeneous at visual inspection (see Figure 3.3), yet it is expected to present an internal structure and two layers with different mechanical and optical properties.

3.3 Results and discussion: To demonstrate the working principle of the probe, we surveyed the sample by performing B-scans in different locations, to identify the position of the underlying step structure. Then, we included that area-of-interest in a C-scan of 3x3mm and eventually performed mechanical indentation along a B-scan. In the OCT image, the structure of the sample is clearly recognizable, as shown in the B-scan in Figure 3.4. Backreflections from the sensor are noticeable at fixed vertical positions in the upper part of the image, but they are not intense enough to jeopardize image quality. Such backreflections are unavoidable in this type of sensor, as they arise from the mismatch of the indexes of refraction at the four interfaces of the sensor: starting from the top, we identify the fiber-to-air, the airto-cantilever, the cantilever-to-sapphire-lens and the sapphire-lens-to-air interfaces. Measuring in a liquid environment would mitigate the mismatches and the intensity of those backreflections. However, the hydrodynamic forces on the cantilever would then impose an upper limit on the scanning velocity. In air, on the contrary, we can drive the stage at full speed and obtain a B-scan in 40 ms and a C-scan in less than 3 minutes. In Figure 3.4 the contrast between the two layers is due to their different scattering coefficient, tuned to be lower in the top one. It is possible to see a smearing of the defect, resulting in a smoothened transition between the region with a thin top layer and the region with a thicker one. The upper surface is smooth but not perfectly horizontal, due to imperfections in the fabrication procedure. After imaging, we performed mechanical measurements across the defect. Specifically, we measured the load required to reach an indentation depth of 15 73

Chapter 3 — Multimodal probe for OCT and indentation

μm, and repeated that every 150 μm along the whole 3 mm length of the B-scan. The results are reported in Figure 3.5. In first approximation, the observed mechanical behavior depends on all the material included in a radius of around 10 times the indentation depth, in our case extending to around 150 μm: this is the depth until which mechanical properties are “sensed”. In the region where the harder layer comprises a smaller fraction of such volume, the load to reach 15 μm depth is therefore lower than the region in which the top and harder layer is thicker (see Figure 3.5). The calibration procedure, from which we obtained a stiffness of the cantilever of 21.3±0.6 N/m, is the predominant source of error on the load (see error bars in Figure 3.5). In these measurements, our ability to recognize two areas and ascribe the different observed mechanical behavior to the respective subsurface structures is a direct consequence of the way our sample was created. In fact, the mechanically stiffer PDMS layers presents the lesser scattering coefficient: optical contrast and mechanical contrast have been tailored to have an identical spatial distribution. However, in some samples, such as breast cancer(K. M. Kennedy et al. 2013), the sources of mechanical contrast might be different from the sources of scattering-based optical contrast. In those cases, performing independent imaging and mechanical measurements as done in this paper would not to be as insightful as demonstrated in this work. This leave room for future developments, aimed at a more intimate integration of OCT imaging and mechanical measurements, with the implementation of OCE techniques. One major hurdle would be the proper quantification of stress within the sample: most current OCE protocols apply the mechanical stimulus over the whole imaging area, and proceed with the analysis in the assumption of a uniform stress distribution inside the material. Our probe does not require precompression, and exerts loads smaller and more confined than any other quasi-static OCE probe(Mulligan et al. 2016). While this is desirable in many life sciences applications, it would not allow us to assume that the stress is uniform along the axis of indentation for the whole imaging range. The depth-resolved quantification of mechanical properties would therefore require more refined modeling of the propagation of the stress field in the material upon indentation.

Results and discussion: â&#x20AC;&#x201D; 3.3

Figure 3.4: B-scan acquired with the sensor hovering on top of the sample. The total horizontal length is 3mm, the vertical length is a 5mm optical-path-difference, and it has been rescaled to represent the true (physical) size of the PDMS sample. The material of each layer has been labeled, the horizontal lines in the upper part are the backreflections happening at the interfaces of the components of the sensor.

Chapter 3 — Multimodal probe for OCT and indentation

Figure 3.5: Load applied to reach 15μm indentation depth, superimposed on the B-scan along which the measurements took place. The error bars are due to the error on the spring constant of the cantilever. The 21 measurements are 150 μm apart, and they cover the 3 mm length of the B-scan. Load is calculated by multiplying the deflection of the cantilever by its spring constant, quantities which are respectively measured by Fabri-Perot interferometry and by a calibration procedure.

3.4 Conclusions: We showed the working principle of a multimodal probe that combines indentation at micro-scale with OCT in an all-optical epidetection scheme. With this sensor, it is possible to serve two purposes: first, one can hover on a structured sample and identify a morphological area-of-interest via OCT, and second, one can perform localized and accurate mechanical measurements in that area-of-interest. The epidetection scheme offers the added value of being able to image thick samples, paving the way to a novel quantitative approach to optomechanical investigation of non-transparent, thick tissues.

Towards clinical elastography of dermal tissues: a medical device to probe skin's elasticity through suction, with subsurface imaging via Optical Coherence Tomography

Abstract The mechanical behavior of dermal tissues is unarguably recognized for its diagnostic ability, and in the last decades received a steadily increasing interest in dermatology practices. Among the various methods to investigate the mechanics of skin in clinical environments, suction-based ones are especially noteworthy, thanks to their qualities of minimal invasiveness, and relative simplicity of setups and data analysis. In such experiments, a structural visualization of the sample is highly desirable, both in its own right, and because it enables elastography. The latter is a technique that combines the knowledge of an applied mechanical stimulus and the visualization of the induced deformation to result in a spatially-resolved map of the mechanical properties, which is particularly important for an inhomogeneous and layered material such as skin. We present a device, designed for clinical trials in dermatology practices, that uses a handheld probe to (1) deliver a suction-based, controlled mechanical stimulus, and (2) visualize the subsurface structure via through Optical Coherence Tomography. We also present a device-agnostic data-analysis framework, consisting of a Python library, released in the public domain. We show the working principle of the setup on a polymeric model, and on a volunteerâ&#x20AC;&#x2122;s skin.

This Chapter has been accepted for publication: Bartolini, L., Feroldi, F., Slaman, M., Weda, J. J. A., De Boer, J. F., van Zuijlen, P., & Iannuzzi, D. (2020). Towards clinical elastography of dermal tissues: a medical device to probe skin's elasticity through suction, with subsurface imaging via Optical Coherence Tomography. Review of Scientific Instruments.

Chapter 4 — Towards clinical elastography of dermal tissues

4.1 Introduction The interest in the mechanical characterization of dermal tissue has been steadily growing in the last decades(Guy 1996, Draaijers et al. 2004, Brusselaers et al. 2010, Berardesca et al. 2014) and its relevance is widely recognized in several fields. Dermal mechanical properties vary with age(Escoffier et al. 1989), location, hydration and other factors(Firooz et al. 2012, Corr & Hart 2013); they are useful as diagnostic indicator to evaluate therapeutic strategies, investigate wound healing, classify burn scars(Draaijers et al. 2004, Brusselaers et al. 2010), assess quality of artificial skin substitutes(Ryssel et al. 2010, Nicoletti et al. 2015), and for the placement of incision lines in excisions and surgeries(Ní Annaidh et al. 2012). In cosmetics, the quantification of skin’s mechanical properties offers a significant help in the substantiation of claims from the industry. Dermal mechanics is also relevant in biomedical applications such as haptics(Bartels 2011), and in pathological disorders such as collagen diseases(Pedersen et al. 2003). The demand to reliably assess the skin’s mechanical behavior has been triggering the development of a plethora of tools to measure it in many ways. At the tissue-level (mm to cm scale), the most common characterization methods can be divided in five major categories: tensile, torsional, indentation, impact (elastic waves propagation), and suction(Rodrigues 2001, Pensalfini et al. 2018). Less common approaches include reviscometry, tonometry, adherence and quantitative electrical characterization(Graham et al. 2019). Among all these techniques, suction-based ones are among the most widespread, due to their in-vivo applicability and to the simplicity of their design. A small negative pressure is created in a chamber and applied to the skin via an aperture. The suction causes the skin to be drawn inwards, and the resulting deformation is used to quantify the observed mechanical behavior. Most academic studies have been making use of custom-made research devices; yet, since its introduction, several groups have been investigating the mechanical properties of skin with the aid of commercial instruments, namely, the Cutometer® (Courage and Khazaka Electronic GmbH, Köln, Germany), the DermaFlex elasticity probes (Cortex Technology, Hadsund, Denmark), and, more recently, the Cutiscan (Courage and Khazaka Electronic GmbH). The Cutometer and the DermaFlex can only measure indirectly the maximum height reached by the drawn-in tissue, while the Cutiscan can take images of the deformation of the whole surface. All of them, however, lack a fundamental feature, needed to fully understand the mechanical behavior of the sample analyzed: they cannot visualize subsurface deformations and structures. Yet, skin 78

Introduction — 4.1

is a complex, multilayer, non-linear and anisotropic material(Joodaki & Panzer 2018): the possibility to image its structure and see what happens under its surface during a mechanical stimulus is of crucial importance in the quantitative assessment of its mechanical properties(Fong et al. 1997, Vogt & Ermert 2005, Hendriks et al. 2006). As put by Graham et al. in a recent review(Graham et al. 2019), “the difficulty in relating the contribution of complex and variable skin structures to the averaged mechanical behavior of a poorly defined skin volume have hampered progress in understanding skin mechanics”. In another review, Jor et al. (Jor et al. 2013) say “there is a need to reliably identify often a large number of unknown model parameters. This is critically dependent on the availability of experimental data that cover an extensive range of different deformation modes, and quantification of internal structural features”. These considerations expose a compelling need for the visualization of the tissue below its surface, and the tracking of its deformation during a suction experiment. That information would offer better insights into the observed mechanical behavior, and would allow for the implementation of mechanical models that consider the additional observables, such as the thickness of the different layers, their collagen density and alignment(Jaspers et al. 2017), and local structures or inhomogeneities, such as follicles, vascularization, or scarring. The integration of subsurface imaging into a suction device was presented in the past by Diridollou et al. and Vogt et al. (Diridollou et al. 1998, Vogt & Ermert 2005), who used handheld ultrasound, and by Hendriks et al., Delalleau et al.(Alexandre Delalleau et al. 2006, A. Delalleau et al. 2008) and Zheng et al. (Hendriks et al. 2004, 2006, Zheng et al. 2014), who used tabletop Optical Coherence Tomography (OCT). These devices provide significantly more information with respect to the commercial ones, as they can measure the profile of a cross-section of skin under suction, visualize its subsurface structures, and quantify the thickness and the deformation of each layer. For example, Vogt et al., produced strain elastograms (i.e. images that map the strain intensity), whereas Hendriks et al. and Delalleau et al. used the additional information in inverse Finite Elements Analyses. A disadvantage of ultrasound-based suction devices is that they need to make use of suction chambers partially filled with water, which acts as coupling medium between the ultrasound transducer and the skin. However, the need to avoid water spills during operation makes the preparation and the measurements somewhat cumbersome to perform. Furthermore, the presence of an intervening medium poses limitations on the ease and speed of disinfection necessary for clinical use. As final shortcoming of these devices, it is worth noting that to obtain 2D cross sections (Ultrasound B mode) the transducer must be mechanically scanned within the suction chamber, a detail that severely 79

Chapter 4 â&#x20AC;&#x201D; Towards clinical elastography of dermal tissues

complicates the engineering requirements. To obtain a 3D images, the setup would need two orthogonal scanning axes within the probe, which would even further complicate the design. The devices presented by Hendriks et al.(Hendriks et al. 2004, 2006) and Zheng et al. (Zheng et al. 2014) integrate Optical Coherence Tomography (OCT) as contactless imaging tool, to obviate the shortcomings of ultrasound imaging, allowing for 2D imaging in air, with a higher resolution. The measurements obtained in those studies were milestones in a better understanding of the mechanical behavior of skin and its modeling. However, they are tabletop devices, so they cannot address the needs of dermatologists, pathologists and cosmetologists to perform studies in clinical environments. Inspired by the recent developments in this area, we have started a project to develop a clinics-ready device to measure the mechanical behavior of skin under suction, with depth-resolved imaging by OCT. Our device is designed to comply with the level of safety required for an investigational device, it is portable, and it makes use of a disinfectable handheld probe with which an operator can measure skin in any location. Its design is adapted from an imaging probe that was developed by some of the authors(Jaspers et al. 2017) and used in clinical studies. In the design process of the setup, we kept into consideration means of patientand operator-protection, portability and versatility. With respect to the aforementioned tabletop devices for skin suction, our handheld OCT offers an improvement of the resolution, and the demonstrated ability to perform 3D imaging. In this paper, we present our setup, which meets all the requirements for clinical trials, and our data analysis framework, which was used to process data from two types of samples: a polymeric skin model and in-vivo real skin.

4.2 Materials and Methods 4.2.1

Experimental Setup

The device presented in this work assesses the mechanical behavior of dermal tissues through the suction method, via a handheld probe, in a portable setup ready for clinical settings (see Figure 4.1). The handheld probe is tethered by a two-meter long cable which allows one to comfortably reach any desired location. The probe tip hosts the small suction chamber with a circular aperture on the outer side. To begin a measurement, the operator brings that facet in contact with the research subjectâ&#x20AC;&#x2122;s skin; partial 80

Materials and Methods â&#x20AC;&#x201D; 4.2

vacuum is produced in the chamber, causing the suction that pulls tissues inwards through the aperture. Simultaneously, the OCT imaging system images the aperture at video-rate, so that the resulting deformations can be spatially resolved, and their evolution tracked. By knowing the applied mechanical stimulus (the magnitude of the suction) and those deformations, one can later model, in principle, the observed mechanical behavior and obtain mechanical properties. The setup consists of the following 4 subsystems: A. Handheld probe, with suction chamber and components for OCT: The handheld probe, whose tip is designed to get in contact to the patientâ&#x20AC;&#x2122;s skin (see Figure 4.1(b)), is connected to the rest of the setup by a two-meter-long flexible bundle which hosts the vacuum tube, the optical fiber and the electrical cables for the scanning galvanometric mirrors. The external casing is made of Polyoxymethylene (POMC), a polymer that allows for easy disinfection while offering excellent chemical and mechanical resistance. It encloses a baseplate on which the free-space optomechanical components for the OCT imaging are mounted. Specifically, light for the OCT is carried by an SMF-28 optical fiber, which terminates into a single-lens collimator (F280APC-C, Thorlabs GmbH). The collimated beam is reflected on two perpendicular galvanometric mirrors (6210H, Cambridge Instruments) that steer it onto a telecentric objective (LSM03 Thorlabs GmbH) and ultimately into the sample. As in a typical OCT system, the light backscattered from the sample returns into the fiber following the same, reversed, path. The suction chamber is at the front-end of the handheld probe and is a twocomponent POMC element in the shape of a truncated cone, whose axis is collinear to the OCT beam. Its wider base, fixed in front of the objective, has a slot to mount an antireflective N-BK7 window (Thorlabs GmbH), which allows for the transmission of the light for the OCT while also sealing the chamber on its side. This base-component of the suction chamber is equipped with an interchangeable cap which hosts the aperture through which the skin is draw-in during the measurements. The cap can be switched to allow for different aperture sizes, varying from 4mm to 12mm. The aperture size is crucial for a multiscale analysis of skin: smaller apertures receive a bigger contribution from its upper portion, whereas larger ones allow for the mechanical stimulus to reach further into the depths of the dermal tissue(Hendriks et al. 2006). B. Optics module: To generate and measure the OCT signal, we use a tabletop AXSUN OCT Engine (Axsun Technologies, USA): an OEM device that includes the processing unit which returns complete B-scans. The system relies on a class-I 30 81

Chapter 4 — Towards clinical elastography of dermal tissues

mW swept-source laser at 1310 nm with a wavelength tuning range of 140nm. We measured its resolution as the full-width-half-maximum of the peak resulting from the reflection of a mirror in air, obtaining an axial resolution of 10.4 μm. The native A-scan rate is 100 kHz, but, due to Ethernet data-transfer requirements, it must be reduced to one eighth, resulting in an A-scan rate of 12.5 kHz. C. Vacuum line to the elastometer: A vacuum pump (10KD series, Boxer GmbH), with a rated maximum negative pressure of 500 mbar (below atmospheric pressure), creates the partial vacuum in the suction chamber. A flexible hose with a 6 mm diameter connects it to the suction chamber. As extra safety-measure, we added a mechanical valve that can be promptly activated to manually vent the suction chamber in case the patient experiences pain or discomfort during the test. D. PC and Software: The setup is run by a PC unit with dedicated software (Labview, NI Instruments) to control each subsystem, and acquire and process measurements. A python library (available as supplementary material and on GitHub, see Data Analysis section) allows for the post-processing, analysis and plotting of the acquired data. The setup, shown in Figure 4.1(a) has been designed with the aim of providing a medical device to clinicians, and was built to match the specifications and standards required for the Medical-Ethical approval. It is mounted on a medical cart (Jannsen Medicars) which conforms to the mechanical and electrical safety principles defined in the IEC 60601 series of standards.

4.2.2

Sample preparation

The device was tested in-vivo on the skin of a healthy volunteer, and on a polymeric skin-phantom. The phantom models the skin structure, and it is made of PDMS (Polydimethylsiloxane, Sylgard 184, Dow Corning). To produce it, two liquid components, the elastomer and the curing agent, were mixed in appropriate ratios, degassed and cured overnight at 60°C. The phantom consists of two PDMS layers – a stiffer one on top and a softer one as substrate – mimicking respectively the epidermis and the dermis. The stiffness of each layer was tuned by varying the ratio of curing agent to elastomer (1:20 for the top one and 1:40 for the bottom one). The top layer was structured to have two regions with different thicknesses, with the transition between them made as sharp as possible (see Figure 4.2).

Materials and Methods â&#x20AC;&#x201D; 4.2

Figure 4.1: Medical device. (a) Photo of the whole setup on its medical cart; the handheld probe is highlighted in light blue. (b) Enlarged 3D design of the handheld probe. It is a custom design, designed and machined by the VU University of Amsterdam mechanical workshop. To improve legibility, the following components have not been displayed: the upper half of the handheld shell, the cables that pass through the strain-relief connector at the bottom of the grip (i.e. the optical fiber for the OCT and the electrical wires driving the galvanometric mirrors), and the flexible hose that brings partial vacuum to the suction chamber.

In PDMS, different curing ratios do not offer the optical contrast needed for OCT visualization: so we mixed in each layer a different amount of scattering agent (TiO2 particles). In this way, the two layers differ in mechanical properties as well as optical ones, and the mechanical contrast is co-located with the optical one, as already described in previous works(de Bruin et al. 2010, Bartolini et al. 2017, Jelvehgaran et al. 2017). The measurement on in-vivo skin was performed on the thumb of a volunteer.

4.2.3

Data acquisition

The software for data acquisition is written in Labview 2016 (National Instruments). It controls 2D or 3D OCT imaging during the application of the partial vacuum. For the former, we only oscillate the fast-scanning galvanometric mirror to obtain cross-sectional tomograms of the sample (B-scans) at a rate of 30Hz; for 3D imaging we also oscillate the second galvanometric mirror, obtaining C-scans (volumetric stacks of B-scans) at a slower acquisition rate of around 0.33 Hz.

Chapter 4 — Towards clinical elastography of dermal tissues

The instrument can be used in two different data acquisition modalities: dynamic or static. For the dynamic one, we continuously record B-scans of the sample for the whole duration of the measurement. This modality is used to investigate the time-dependent (viscoelastic) behavior of skin. For static analysis, instead, we consider only the elastic component of the behavior and ignore the transient status, recording the sample only in the unloaded condition and the one at equilibrium under suction. The static modality is sufficiently slow to allow for C-scan recording, whereas in dynamic mode we are restricted to 2D imaging. In the following proof-of-concept measurements, the pump acted at full power, causing the pressure in the suction chamber to drop by 500mbar as quickly as possible; the suction was then removed by manually venting the chamber.

4.2.4

Data analysis

The ad-hoc software for the data analysis is written in Python, and it is publicly available at https://github.com/LucaBartolini/OCT with a MIT license. This public repository contains sample data, example scripts (Jupyter Notebooks) to analyze them, a helper tool to create arbitrary suction profiles, and the library at the core: “OCT_lib”. It is important to stress that, while the library has been designed to work with the specific hardware presented above, it can be used with arbitrary input images (from other OCT setups, or even other techniques). The library has been designed from scratch, to provide an organic, user-friendly API framework (Application Programming Interface) to post-process the OCT data from a suction experiment. It is responsible for the readout of the OCT data, the preprocessing and filtering of the images, their segmentation to find the surface, and finally their plotting. The library provides a high degree of customization of the parameters used in the data analysis, while also offering default settings to work out-of-the-box. Its most used feature in this work is the automatic surface recognition via image segmentation. With that algorithm, we measure the height “z” of the skin’s surface along the B-scan position “ ” and obtain a 2D profile …( ). We call deformation profile Δ…( ) the height difference between a profile and the profile in unloaded condition. In other words, Δ…( ) represents the mechanical deformation across the aperture.

Results — 4.3

4.3 Results This section presents experimental result from dynamic and static tests, with the intention to showcase the range of possibilities and amount of information provided by our device. In order, we will present the static measurements on the phantom, introduce a descriptor for the deformation profiles, display volumetric images on skin, and finally we will show a dynamic test on skin.

4.3.1

Static measurement on phantom, asymmetry factor

As first proof-of-concept for a static analysis, we imaged the PDMS skin-phantom at two equilibrium conditions: unloaded, and at 500mbar suction. Our API identified the surface, and the calculated the deformation profile, illustrated in Figure 4.2. On the phantom sample, we can identify three regions, namely “thick”, “thin” or “boundary”, which respectively comprise parts of the PDMS sample with only the thicker top-layer, only the thinner one, or the boundary region which contains the sharp transition between the two (see Figure 4.2(a)). The possibility to spatially resolve the surface deformation along the whole aperture can be very valuable in preclinical studies on skin. However, its interpretation in a clinical setting could be too demanding, since it undoubtedly requires more cognitive resources and subjective judgement. Therefore, with the aim of summarizing the information of the 2D deformation profile, we introduce the Asymmetry Factor (AF), which captures one of the meaningful features of the profile. In fact, given that the partial vacuum induces a uniform loading condition, a deformation profile exhibiting a certain degree of asymmetry indicates inhomogeneities in the structure or mechanical properties of the sample. In the case of skin, for example, such asymmetry could suggest the presence of scar boundaries and/or collagen-rich structures. To demonstrate this concept, we have performed measurements on the PDMS skin-phantom in the three regions defined above, expecting to notice a clear asymmetric mechanical behavior in correspondence of the boundary region, where the top layer goes from high to low thickness. From a given profile deformation Δz( ), we take only data that are above 20% of the peak height to eliminate boundary effects. We divide the base of the new profile in two halves and call the corresponding profiles … ‡ˆ and … 5 . Finally, in Equation 4.1, we define the AF as the ratio of the difference of the norms over their sum: =

‰… ‰…

‡ˆ ‡ˆ

‰ − ‰… ‰ + ‰…

5 5

‰ ‰

(4.1)

Chapter 4 — Towards clinical elastography of dermal tissues

where, ‖…‖ = 9∑ …( ). , the common L2-norm. This makes sure that the AF is 0 when the profile is symmetric and that it increases with the asymmetry of the profile up to a maximum of 1 for antisymmetric profiles. As for its sign, AF is positive for a profile bulging to the right and negative for a profile bulging to the left. We measured each PDMS region (thin, thick, and boundary) five times and calculated the AF for each of them. The results are shown in Figure 4.3. Figure 4.3(b) shows a significant difference between the boundary region, where the thickness of the top layer varies along the profile, and the two regions thin and thick, where the thickness of the top layer is homogeneous.

4.3.2

Static measurement on skin

We have also performed static test on the fingertip of a volunteer’s thumb. Figure 4.4shows two B-scans of that sample at equilibrium in two different conditions: unloaded, and at 500mbar suction. A few additional datasets of static measurements on different locations (i.e. forearm, backhand) are provided as supplementary material – available in the folder “02-Example-data” in our public repository: https://github.com/LucaBartolini/OCT/ By definition, the deformation of the sample at equilibrium is stationary, so we had the time to acquire a C-scan: this volumetric image is shown in Figure 4.5. The C-scans in Figure 4.5, made of 290 B-scans, took approximately 3 seconds each. After a preprocessing of the B-scans via OCT_lib, the 3D composition was done with the help of imageJ.

Results — 4.3

Figure 4.2: Static measurement on the PDMS phantom, on a single location in the “boundary” region. On the PDMS sample, three regions are identified, namely: “thick” (where the top layer is thicker), “thin” (where the top layer is thinner), or “boundary” (the region of transition between “thick” and “thin”). The surface profile, as detected by the segmentation algorithm in OCT_lib, is highlighted in red. (a) OCT image of the sample in unloaded condition; (b) Same as (a) after the application of 500mbar suction. (c) Definition of “deformation profile” $…( ): the point-by-point difference of the two profiles along the whole aperture. (d) A smoothed deformation profile, highlighting T ‡ˆ and T 5 , which will be used in the calculation of the asymmetry factor (AF). The deformation profile is centered, so that its peak is found at position 0 on the abscissa

Chapter 4 — Towards clinical elastography of dermal tissues

Figure 4.3: Summary of all measurements on the PDMS skin-phantom: five different locations in three different regions. (a)Profile deformation Δz(x) of each single measurement, smoothed by a Savitzky–Golay filter. (b) Avg. and standard error of the mean of the AF of profiles in (a)

Figure 4.4: B-scan of a volunteer’s left thumb fingertip. The sloping walls of the suction chamber are visible on the sides. The surface profile, as detected by the segmentation algorithm in OCT_lib is highlighted in red. (a) B-scan, unloaded condition. (b) C-scan at equilibrium, under a suction of 500mbar.

Results — 4.3

Figure 4.5: C-scan of a volunteer’s left thumb fingertip. The sloping walls of the suction chamber are visible on the sides. A reflection artefact appears as an ellipsoid hovering over the surface. (a) C-scan, unloaded condition. (b) C-scan at equilibrium, under a suction of 500mbar.

4.3.3

Dynamic measurement on skin

Figure 4.6 shows an example of what dynamic measurements are possible. In dynamic tests, we observe the time series of the deformation profiles of the drawn-in skin during suction. This approach is similar to the one used in commercially available instruments. However, most clinical devices can only track the maximum height of the deformation – a feature that we can easily reproduce (see Figure 4.6(a)). On that time-series one can identify salient points, and use them to obtain mechanical descriptors such as the Cutometer-specific “Rparameters” (stretchability, fatigue, net elasticity, etc.), or other parameters from more advanced constitutive models (Hendriks et al. 2003, 2006, A. Delalleau et al. 2008). Our instrument can follow the evolution in time of the entire 2D deformation profile Δ…( , S), as shown in Figure 4.6(b). Those profiles are obtained by acquiring B-scans in the same location, at a frequency of 33.3 Hz, for the whole duration of the test, as partial vacuum is created and then released. We take as reference unloaded profile the average of the first 10 surface profiles and subtract it from each following profile to obtain the deformation Δ….

Measurement error

The error on the deformation profile Δ…( ) has two major sources: the resolution of the OCT, and the inaccuracy of the surface-recognition algorithm. 89

Chapter 4 — Towards clinical elastography of dermal tissues

The total error will be given by the sum in quadrature of the error form each source.

The resolution of the OCT contributes to the error with …Œ•Ž = 10.4 jk. Since the real position of the surface is unknown, the estimation error of the algorithm to find its position is unknown. However, we can assume an inaccuracy of 2 pixels, to add each time it is applied, once for the reference and once for the loaded surface. Every OCT system slightly oversamples the axial direction so that the physical dimension of each pixel is slightly under the resolution; in our case, each pixel is … : = 5.9 jk. Therefore, the total error Δ… on the estimation of the deformation profile Δ…, is: Δ… = • …Œ•Ž . + 4 ⋅ …

≈ 16 jk

(4.2)

4.4 Discussion The PDMS phantom in Figure 4.2 was purposely created to display co-localized mechanical and optical contrast. Thanks to this quality, we could identify mechanical inhomogeneities (the step-change in top-layer thickness) by OCT imaging alone. However, the two sources of contrast are not necessarily associated, and the AF measurements shown in Figure 4.3(b) could prove a good indicator of underlying boundaries or structures, in case the optical contrast would fail to do so. It is worth noting that if one were to closely observe the profiles in Figure 4.3(a), they could identify a degree of asymmetry in the profiles of the “boundary” region, but the AF summarizes and captures that in much clearer way. With the B-scans shown in Figure 4.4 we can visualize the skin’s structure indepth, clearly distinguish the epidermis and dermis, and possibly more structures. That sort of visualization is an extremely valuable diagnostic tools per-se, but it also allows one to extract parameters for constitutive modeling, such as epidermal thickness, or investigate the relation between skin structures and mechanical behavior. Data from static measurements can be used to obtain the mechanical properties of skin under suction, either through analytical models, as shown by Diridollou(Diridollou et al. 2000), finite element modeling(Khatyr et al. 2006), or borrowed with caution from micropipetaspiration theories(Lim et al. 2006).

Discussion â&#x20AC;&#x201D; 4.4

Figure 4.6: Dynamic analysis of the deformation under application of a 500 mbar suction to a volunteerâ&#x20AC;&#x2122;s thumb fingertip. (a) Time series of the maximum of the deformation profile, reminiscent of the commercial CutometerÂŽ data acquisition. After a short initial delay, a 500 mbar suction is applied to the sample, and later removed. (b) Time series of the complete deformation profiles. Highlighted in red, is the maximum deformation for each profile: those points are the source data for the graph reported in(a). Colors represent the relative height with respect to the maximum deformation for each profile.

Chapter 4 — Towards clinical elastography of dermal tissues

In Figure 4.5 we show whole C-scans, before and after the application of 500 mbar suction. No quantitative analysis has been carried on those data, yet the qualitative information they provide is appreciable by itself. For example, the shape of the deformation surface could be used to infer Langer Lines: the lines of preferred natural tension in the skin. In fact, they would reveal themselves as an azimuthal anisotropy, i.e. the surface curvature would change along the different planes passing by the “axis” of the suction. Figure 4.6 shows an example of dynamic measurement. On one hand, we can imitate the data acquisition modality of commercial devices (Figure 4.6(a)), providing a base for the validation and comparison of our device to others. On the other hand, we can obtain more information than what was previously possible as in Figure 4.6(b). The visualization the whole deformation profile and its time evolution could provide new insights into the mechanics of skin. For example, we can observe that the ridges and valleys of the fingertip experience a different strain. The ridges are visible both in the unloaded state and in the loaded one. Yet, the deformation profile Δ… shown in Figure 4.2(b) is the difference between the two states and it still displays the same pattern, indicating that valleys and ridges undergo different absolute deformations. This could be of interest, for example, in the modeling of fingertips, for haptics design(Leonardo 2010, Miguel et al. 2015) and tactile sense studies. In Figure 4.6(b), moreover, one can see how the position of the maximum of the deformation profile (the red dot) is not always consistent, but jitters, indicating that the peak considerably extends to its sides. Finally, another feature that we can only appreciate in those 2D profiles is the “flatness” of the peak: the same maximum deformation could originate from a sharp profile, or from a “flat-top” one. This quality could be of interest as at-aglance descriptor, so we chose to illustrate it by coloring the surface: while the shape of the surface represents the absolute deformation Δ…, its color represents the deformation relative to the maximum, i.e.

•‘

’“” (•‘)

for that specific frame.

Samples with a sharp “peak” in the deformation profile would show a higher color gradient, whereas flat-tops like in the example present a more uniform coloring around the maximum. Along the line of our vacuum subsystem was mounted a closed-loop pressure controller (EL-PRESS P-800, Bronkhorst Nederland B.V.), to allow the user to predefine an arbitrary history (or fixed setpoint) of negative pressure, limited only by the pump specifications for maximum negative pressure and flow rate. Yet, as explained in “Data Acquisition”, in the showcased measurements, the pressure regulator was bypassed to allow the suction chamber to reach an absolute pressure of 500mbar as quickly as possible, resulting in only two possible loading conditions (either no-or maximum-load). Further studies could 92

Conclusions â&#x20AC;&#x201D; 4.5

employ the full potential of the pressure controller, for instance by applying suction in step-wise increases to test the non-linearity of skin(Hendriks et al. 2003, A. Delalleau et al. 2008).

4.5 Conclusions We presented a device that, thanks to a handheld probe, can measure the mechanical behavior and visualize the structure of dermal tissues non-invasively, in-vivo, in preclinical and clinical environments. We also presented a Python library that is now in the public domain (https://github.com/LucaBartolini/OCT) that processes and analyses the obtained data, offering a framework to get new insights into the mechanical behavior of skin, and which can be used more in general in other suction-and-imaging experiments. With our device, we have shown static and dynamic measurements. Static test disregard transient mechanical behavior, but they provide us with the time to obtain a 3D depthresolved image of skin. Dynamic tests investigate the viscoelasticity of the dermal tissues, but tracking the temporal evolution of the deformation profile, requires a minimum imaging frequency that is achievable only in 2D. We have proposed a quantity named Asymmetry Factor, that could be used to indicate mechanical inhomogeneities of the sample. Finally, we have shown the possibility to acquire 3D tomograms.

5 Valorization

The following Chapter is a â&#x20AC;&#x153;gedankenexperimentâ&#x20AC;? in entrepreneurial strategy. It addresses the essential elements of a business plan to commercialize the device that was built and developed during the course of my PhD. It is a learning exercise to apply fundamentals of (innovation-) business development to this specific use case, with the awareness that the full depth of an actual business plan is not within the scope of this work. Here, I imagine teaming up with my supervisor, Prof. Davide Iannuzzi, setting a strategy, and preparing the ground for a spin-off company that brings the outcome of my research project into the research- and medical-devices market.

Chapter 5 — Valorization

5.1 Introduction The medical-research device discussed in detail in the previous Chapter 4, has two main functions: (1) it measures the mechanical response of dermal tissue under suction, and (2) it visualizes its internal structure, and the deformation caused by the suction, via a depth-resolved imaging technique called Optical Coherence Tomography. The combination of mechanical stimulation and depthresolved imaging of the induced deformations enables one to map the spatial distribution of the mechanical properties in the cross-section. Dermal structures and their mechanical properties have a widely recognized diagnostic power, applied in many fields, including the assessment of scars and healing therapies or quality control of artificial substitutes. Our device is currently at a proof-ofconcept stage (i.e. Technology Readiness Level (TRL) between 3 and 4, according to European Guidelines(European Commission 2015)), and was built following suggestions from our medical collaborators at the Beverwijk Burn Center, with the intention of eventually using it in their clinic. The formal application for clinical trials is mostly completed at the time of this writing.

5.1.1

Business Model Canvas

The Business Model Canvas (BMC) was proposed by Osterwalder in 2006 to display the strategic information that constitutes a business plan into a concise visual arrangement. In the past decades, it has become increasingly clear that traditional business plans were a sub-optimal tool for dynamic, innovative startups. Indeed, it is recognized that “a business plan never survives first contact with the customer”(Blank & Dorf 2013): spending weeks on a carefully constructed 50-page business plan is now recognized inefficient at best, and damaging due to Confirmation Bias at worst. A solid alternative was thus recognized in the BMC, which consists of a single-page layout of nine building blocks covering the essential elements of a business plan: the product-market fit, the company (infra)structure, and its finance (see Figure 5.1). Several reasons contributed to the diffusion of the BMC. To start with, it is a framework to analyze the business strategy in a systematic way, making sure that no important question is neglected. At the same time, its minimalistic format allows for a certain degree of front-end fuzziness, an increasingly valued quality in New Product Development(Brun 2012). The BMC’s compact format also helps to provide clarity by focusing only on the core drivers of the business and leaving out what, at the early stage of a business, is superfluous. Finally, and perhaps most importantly, its conciseness is valued by itself: all the information is displayed at once, it can be written down and changed quickly as new information comes in, and above all, it can be read at a glance. 96

Introduction â&#x20AC;&#x201D; 5.1

Figure 5.1: Building blocks of the Business Model Canvas. Reproduced with permission from: Osterwalder, Pigneur, Yves - Business Model Generation(Alexander Osterwalder 2010)

Owing to all these qualities, it is particularly widespread within the innovation/technology business, as it appeals to all stakeholders: entrepreneurs, investors, and teams, who are concerned with either designing, improving, understanding, discussing or pivoting the business strategy. Recently, it has found its way even in the application forms of public funding schemes, becoming an increasingly relevant tool for the scientist alike. Among the major sources of criticism of the BMC is its inability to capture the evolution of the model while presenting only a static snapshot of the business strategy. In this Chapter, this drawback was mitigated, and a dynamic introduced, by identifying two timeframes and writing a Business Canvas for each. Considering the current status of the device and its TRL, the customary 2-5 years projection would be premature: a shorter and more realistic timeframe of two years has been chosen and split. At the end of the Chapter, the following two canvases will be presented: â&#x20AC;˘

a short-term one (zero-to-one), spanning the first 6 months, to codevelop the current proof-of-concept into a Minimum Viable Product (MVP) together with our first customer. That one customer is our collaborator from the Dutch Burn Foundation, plastic-surgeon Dr. Paul 97

Chapter 5 — Valorization

•

5.1.2

van Zuijlen, who initially suggested the features he would want and provided medical oversight to the project; a medium-term one, encompassing the first scaling process (one-to-ten), aimed at selling the device to innovators and early adopters among dermatology practitioners (see Diffusion of Innovation theory(Rogers et al. 2019)). This BMC lays down the strategy for roughly a year and a half, starting after the first sale. The focus is mostly on early business development, product development, and reaching out to those new potential customers.

Effectuation theory

The traditional logic of business planning is based on the determination of a specific goal and of the resources instrumental to reach it. If any specific means to that end is lacking, a great deal of effort is put into obtaining it. Recognizing that entrepreneurial initiatives significantly differ from business strategies developed for established firms, recently, Sarasvathy’s pioneering work (Sarasvathy 2001) defined the concept of effectual thinking, as opposed to the traditional approach named causal thinking. Within effectual thinking, the initial effort is put into a comprehensive internal analysis (codified by the questions “who I am”, “what I know” and “whom I know”) to define and understand all the resources and capabilities currently available to the entrepreneurs. Then, a set of possible effects, or goals, attainable by leveraging on those assets is identified: the entrepreneur (or team) will then choose the most appealing one. Upon reaching that first goal, a better position will be available to understand what skills are desirable, and to interact with and obtain commitment of other stakeholders. They may not only provide input for a good market fit but also, and importantly, all the resources and capabilities that the entrepreneurs may have no direct access to. Then, those fresh assets help defining improved goals, triggering a feedback loop called dynamic model of effectuation. In that loop, the interactions and contingent needs that emerge in the process will bring about expanding resources and converging goals, which in turn will feed back into the cycle. By shifting the focus from the missing assets to what is already possible to do (called bird-in-hand principle in Effectuation Theory), and leveraging on contingencies rather than relying on long term planning, the resource-savvy approach of Effectuation Theory has been shown to be especially effective in the initial phases of decision making in innovation business; once the business matures and goals naturally converge, causal thinking gradually becomes more important, as observed by Berends (Berends et al. 2014).

The four pillars — 5.2

5.2 The four pillars At the basis of any business plan stand four “pillars”: product-market fit, internal analysis, external analysis and value capturing (Iannuzzi 2017). A subsection in this Chapter will be devoted to each of them, with principles of Effectuation Theory supporting several considerations. Eventually, all the information will be reshaped and summarized with the help of the Business-Model Canvas (BMC).

5.2.1

Product-market fit

The Product-market fit (originally dubbed Value Proposition8) is the first pillar, a fundamental concern of any innovative enterprise(Andreesen 2007, Dennehy et al. 2016). It is characterized by three aspects: the added value of the product, its intended (or potential) market, and the way they are related. Its crucial importance is indirectly demonstrated by what happens when it is neglected: among the most common reasons for startup failure, accounting for almost half of them, is the lack of an appropriate market demand (Arnaud 2018, Bednár & Tarišková 2018, CB Insights 2019). Product-market fit is particularly relevant for endeavors of technology transfer from academia, since the design drivers for a research device are not the same as the ones for the private sector. One can easily imagine a setup that researchers find extremely interesting, but with very little market appeal, due to, for instance, high cost, complexity, fragility, or unneeded features (i.e. nice-to-haves that a customer wouldn’t want to pay for). Therefore, it is of utmost importance to thoroughly analyze the product-market fit, and make sure that the proposed added value is addressing a (latent or manifest) problem that someone (the target customer) would pay to solve. Bonsai and tree lovers know very well the duality between a tree’s root system and its canopy: in a healthy tree, one mirrors the other in size and vigor. Something similar can be said for the product-market fit. If the reader allows the metaphor, the leaf canopy of the tree in Figure 5.2 represents the added value of the product: it’s the most visible, and the BMC calls it “Value Proposition”. Below the surface, and equally important, the root system portrays the intended market, which divides into “Customer Segments” (in BMC terms). The trunk is the exchange route: water and minerals are brought to the canopy, where they are processed into sugars and brought back to the roots. Similarly, cash flows from the customer segments into a company, in exchange for a product or service, 8

In his book “Entrepreneurship for Physicists”(Iannuzzi 2017), Iannuzzi names the first pillar “Value Proposition”. In this work, the author chose to call it “Product-Market Fit”, following a different perspective. The focus, rather than being on the inherent features of a product (Value Propositions), is shifted to their contextualization within a marketplace.

Chapter 5 — Valorization

along connections that in BMC terms are named “Channels” and “Customer Relationships”. The centrality of the Product-Market Fit is also reflected by its position in the Canvas, where it is detailed in the four building blocks that were just mentioned. The Value Proposition explicitly states a product’s competitive advantage or added value. To do so, it must address the needs of, and offer gains to, a specific customer segment. In other words, the value proposition must answer the questions of what is offered and why a customer would want that. The Value Proposition is complemented by “Customer Segments”: groups of potential users who share some needs or wants, thus creating a common demand. A “Customer Segment” answers the essential question of who is willing to pay for a specific Value Proposition. As a consequence, the customer segments define several crucial aspects of the business strategy. For instance, they embody the market size, which ultimately sets boundaries to the scalability of the business and to its long-term chances of survival. Specific segments also define the channels to reach customers, the treatment they expect, and the pricing strategies that apply. In practice, their definition (retroactively) shapes the Value Proposition, with the foremost aim of reaching an even better fit. On the canvas, exactly as in real life, value proposition and customer segments are connected by customer relationship and channels. “Customer Relationship” concerns the kind of relationship to be offered to each customer segment, in a spectrum that ranges from dedicated personal assistance to a “help” section on a website. “Channels” puts the focus on the means to reach the segments, show them our Value Proposition, and engage them in a purchase. In our case, the value proposition is already clear, as is our “customer segment”: we developed a clinical research device for in-vivo elastography, with the initial input of our medical collaborator van Zuijlen. He is a plastic surgeon, who has done extensive research on the treatment of burn scars and is therefore interested in the measurement of the elasticity of dermal tissues. It is especially relevant that he pioneered the introduction of suction experiments for the evaluation of the biomechanics of scars, bringing an elastometer device, that was initially developed for the cosmetic world, into the clinics. That elastometer is a reliable machine, whose technology could nowadays be defined as “outdated”. We offered to integrate in a single device: (1) the same working principle of suction experiments on dermal tissues and (2) a modern, depth-resolved, imaging technique.

100

The four pillars â&#x20AC;&#x201D; 5.2

Figure 5.2: A visual metaphor for the Product-Market Fit. Trees are models of exchange systems, and they can represent the duality of the Value Proposition (the treeâ&#x20AC;&#x2122;s canopy) and the Customer Segment (the root system), as well as the connection between them, and their existence in a wider context of market landscape.

Van Zuijlen, in the theory of Diffusion of Innovation(Rogers et al. 2019), would be classified as Innovator, as he is a curiosity-driven person who is willing to take part into high-risk, high-reward endeavors. Our device will appeal to that curiosity, his interest in new technologies, and his desire to bring new instruments into his clinical routines. More importantly, he and his collaborators are already studying the mechanical behavior of skin, doing so by palpation, selfevaluation questionnaires or with the old elastometer. That job of theirs would be improved by a tool that objectively quantifies the mechanical behavior. Moreover, while subsurface structural imaging is already useful by itself, yet its combination with an elastometer would enable a deeper understanding of the observed mechanical behavior and prove an invaluable addition. Finally, the access to a new device will also provide him and his collaborators with the possibility to publish other pioneering papers, a valuable selling point in the academic environment. Customer segments, relationship, and channels have been simplified in this work, and will be discussed in the following Business Canvas sections.

101

Chapter 5 — Valorization

5.2.2

Internal Analysis

The following Internal Analysis finds its roots in “resource-based view of a firm” by Barney(Barney 1991), which provides a framework to identify, categorize and evaluate the assets of a company. He argues that it is especially important to do so, since only worthwhile assets provide a company with sustainable competitive advantage, i.e. the primary source of revenue. Following effectuation theory, and as suggested in (Iannuzzi 2017), the internal analysis for the case here discussed has been performed on the basis of three fundamental questions: “who are the entrepreneurs?”, “what do they know?”, and “whom do they know?”, a comprehensive answer of which would let the entrepreneurs be aware their stance and the possibilities within their reach. A common distinction is made between resources (“what an entrepreneur can count on”), and capabilities (“what they can do with them”); both can be further grouped into 6 categories: human, social, financial, physical (geographical), technological, and organizational (Grant 2000). 5.2.2.1

Resources & Capabilities

Table 1 lists the resources of this project (in rows) and the respective holder (in columns). For each holder, the resource is rated on a scale of one to three dots (· ), whereas a star (*) means the resource is independent of the person.

Table 2 lists the capabilities that those resources enable. In the next section (5.2.3), these assets will be placed in their broader context. Later, in the External Analysis section, we will analyze the value and importance of each of them in relation to the market landscape and competition.

102

The four pillars — 5.2

Table 1: List of Resources LB (author)

DI (supervisor)

Entrepreneurial spirit

···

Eagerness to see the fruit of own work

···

Time to devote

···

Human

Expertise in Lean Startup management

· ··

Social Network of domain-experts in Academy

···

Reputation in Academic World

···

Network of known/trusted suppliers Contacts in private sector

·· ·

··

Access to Research grants

···

Access to VC

··

Incubator spaces in Demonstrator Lab

University's labs and instrumentation

Two Academic medical centers in Ams.

Vivid startup scene in Amsterdam

Netherlands’ policy prioritizes valorization

Financial

Physical

Technological Working prototype Post-processing software

···

Biomechanics knowledge

··

Optics knowledge

··

···

Documentation for Research Medical Device

Clinical trials accounted for

IXA (Amsterdam)

ACE (Amsterdam)

Organizational

103

Chapter 5 â&#x20AC;&#x201D; Valorization

Table 2: List of Capabilities Human Form the right team Build mutual trust with co-developing Medical Dr. Social Interact with Adopters Reach experts to define technological aspects Reach experts to define Product Devt. Process Reach experts to define Business Devt. Interact with other entrepreneurs Financial Be incubated into Demonstrator Lab Apply to research grants Postpone external investments Physical Move the prototype to Academic Hospitals Access the University's laboratories Technological Increase the TRL of the prototype Add features to the post-processing software Quickly develop new designs Work with Mechanical Simulations groups Organizational Rely on Coaching/ Mentoring opportunities Rely on a fast decision-making process Rely on a cost-effective PDP Conduce clinical trials with hospital support

104

The four pillars â&#x20AC;&#x201D; 5.2

5.2.3

External Analysis

The third pillar is External Analysis, in which the aforementioned assets are contextualized, by taking into consideration the size of the intended market, its landscape, and the competition. To gain further insights about the market and learn about the needs/wants of its customer segments, we also interviewed the domain-expert in our network. 5.2.3.1

VRInS Criteria

Once all the assets (resources and capabilities) from the Internal Analysis section are listed, it becomes evident that not all of them are equally important: while some might be indispensable to the success of the company, others might simply be desirable. Their importance is inextricably linked and referenced to external factors: the market and competition settings in which those assets exists. As Barney(Barney 1991) stated, it is imperative to recognize the few, special assets that provide a sustainable competitive advantage, i.e. those resources or capabilities that put an entrepreneur in such a condition that potential competitors are unable to replicate their strategy, either now, or in the coming future. He argues that those assets must fulfill the VRInS criteria: Valuable, Rare, Imperfectly Imitable and Non-Substitutable. While itâ&#x20AC;&#x2122;s somewhat trivial that a reasonable list of assets would only include those which are Valuable, the other criteria deserve special attention. Rare assets are those resources or capabilities that only a few companies can deploy. Imperfectly-Imitable is the quality of an asset to be obtainable only by the holder and no other competing entity, while non-Substitutable means there are no other strategically equivalent assets. VRInS criteria are intentionally stringent: they aim at identifying only the assets that offer the best chance at providing a sustainable competitive advantage. Regarding rarity, it is important to remember that it can be increased by a combination of assets. For instance, in the case of the research device presented in this Chapter, the ability to perform depth-resolved imaging with OCT is not particularly outstanding: there are hundreds of research groups in the world that routinely use this technique, and dozens of companies which commercialize such setups. The study of the mechanical properties of (dermal) tissues through suction experiments is surely less common, but not exceptional. However, the interdisciplinary field between OCT imaging and mechanics is a fertile ground called Optical Coherence Elastography, counting a relatively small number of research groups worldwide. Of those, only a fraction has the knowledge in and ability to bring a research device into the clinics, and even fewer will express any business appetite. Furthermore, the environment our project is developing in, in the center of a vivid entrepreneurial ecosystem that is in close contact with the

105

Chapter 5 — Valorization

innovators of the target market, is also outstanding. In that sense, the medical device, and the assets leveraged to obtain it, meet the rare criteria. The device is not covered by a patent position. Even though there is a great amount of “soft IP” in this project (to be intended as know-how and software development), it is fair to admit that this is probably one of the largest liabilities of the project, as it exposes the technological aspects of the idea-to-market process to the risk of imitation attempts. The other resources and capabilities related to the ecosystem surrounding the project can be, on the contrary, considered difficult to imitate. In terms of substitutability, as argued in the previous Chapters, the combination of suction and OCT does provide a unique way to look at skin. There are no other instruments on the market that can gather the same information that our device can provide, which means that, if others wanted to compete on the same added value proposition, they would most likely need to imitate our offer. Looking at the discussion above, we can conclude that, concerning the resources and capabilities available to the entrepreneurs, one of the strongest assets this project can count on is the environment it is developed in (what Effectuation Theory calls “the crazy quilt”(Sarasvathy 2008)), which seems to be perfectly tailored to the task at hand. The most relevant liability is the lack of a strong IP position, which means that the entrepreneurs will have to design a strategy that could leverage on the surrounding environment to gain a robust first mover position, as discussed later. With this in mind, our imagined idea-to-market process could not be captured better than in the book Engineering a High-Tech Business (Lopez-Higuera, Jose Miguel; Culshaw 2008): “Yes, some capitalize on highly protected scientific breakthroughs, but many more engineer a well-known concept into a particular geographical market or social niche.” 5.2.3.2

Market Landscape and Competition

To understand the competition landscape, we look for companies whose products have similar, or overlapping, value propositions to ours. In this respect, it’s worth repeating that our device serves two purposes: it acts as an elastometer (a diagnostic tool that performs a suction experiment), and as an OCT imaging device for dermal tissues. Notably, their combination also allows for crossdisciplinary features, such as elastography and the study of correlations between structure and mechanics. Regarding the first function, there are already several devices that estimate the (visco-)elasticity of dermal tissue by means of suction: the major players being 106

The four pillars — 5.2

the market leader Cutometer® by Courage & Khazaka, and the DermaFlex® by Cortex Technologies. Those devices come with multiple accessory probes used to estimate different skin parameters, including hydration, pigmentation, and, of course, skin elasticity. They are targeted to the cosmetic industry, but are becoming more and more common in dermatology clinics and research centers. Indeed, clinimetric studies have shown the diagnostic abilities of elastometers(Enomoto et al. 1996, Hristo Dobrev 2014), particularly for burn scars(Verhaegen et al. 2011), and it is safe to say that they are a generally known and established tool in dermatology. The market for OCT imaging is larger and more segmented: as a high-resolution, depth-resolved in-vivo imaging technique, its applications range from ophthalmology (where it’s an established diagnostic tool), to endoscopy, to many other medical and non-medical ones. Consequently, its market landscape is vast and characterized by many players, and its size is currently limited mostly by the high price of an OCT system (50 – 100 k€). However, rather than direct competitors, they can be seen as Original Equipment Manufacturers: companies that build and provide complete OCT solutions to integrate in our system. Conversely, Optical Coherence Elastography, while being a fervent field in the academic world, has not yet found commercial exploitation. However, there is an analogous technique that works at a bigger scale and lower resolution, i.e. Ultrasound elastography, which has already shown clinical success(Shaheen et al. 2007, Afdhal 2012), and reached the market with the Fibroscan® device from Echosens (France). The market landscape just depicted seems to be favorable to our entry, based on a number of considerations: •

•

Elastometers are already known and validated: since our device is proposed as an improved alternative, the switching cost for the original users is greatly reduced, because we can provide the same estimates, while adding (1) spatial information about the induced displacement and (2) structural visualization below the surface Optical Coherence Tomography has already found its way into dermatology, where it is used as a standalone technique, its diffusion limited mainly by its high cost. However, commercial devices offer higher resolution than what would be needed for useful elastography. We could compromise on the resolution, and tap into the developing field of lowcost OCT, to eventually provide subsurface structural visualization at a fraction of the cost of a typical high-resolution OCT device The clinical relevance of elastography at organ-level (Ultrasound Elastography) can be translated to a demand for elastography at tissue107

Chapter 5 — Valorization

•

level, as also widely demonstrated in literature(Khalil et al. 2005, K. M. Kennedy et al. 2015) The working principle of our device has already been successfully shown in the literature(Hendriks et al. 2006, Zheng et al. 2014) before us, but only as a tabletop research device. Our opportunity presents elements of both technology push and market pull. The former is an inherent consequence of the research nature of the project. However, given the points above, and that the development was driven by clinical needs as expressed by our collaborator, we can identify an encouraging degree of market pull.

Therefore, we can conclude that there is an unmet need in the dermatology market, and our device can address it. If that is true, we can assume that we have found a market niche, and that we would be in a favorable position to reap the so-called first-mover advantage: the competitive advantage obtained by being the first entrant in a market segment. Notably, that position is not always advantageous: Suarez and Lanzolla (Suarez & Lanzolla 2005) recognize that it is only so if both the paces of market growth and of technology evolution are relatively slow, a combination dubbed “calm waters”. In fact, they argue, in a market that grows quickly there’s always room for potentially disruptive newcomers; whereas if the underlying technology evolves quickly, the advantage weighs in favor of late-comers, as first-movers invested resources on technologies that might quickly become obsolete. Our case falls relatively safely within their “calm waters” definition. The market is growing at a slow-and-steady, healthy rate (see next section Market Size). The technology of elastometers has been mostly unchanged for decades, hence its pace of technology evolution is slow; whereas OCT technology has seen several recent developments, especially in the direction of low-cost that is relevant for us, and in that sense, we can harvest latecomer advantages. All of this means that we could probably have time to focus on the product development without a considerable risk of being out-engineered or outmarketed any time soon, ultimately building and consolidating our position in the market and turning the first-mover advantage into a sustained one. In fact, that valuable head start will pose barriers-to-entry to latecomers, including the lengthy process to obtain the medical device certifications, the technical challenges, and the pre-emption of resources. 5.2.3.3

Market Size

In our short- and medium-term strategy, the most desirable customers are those geographically closer to us. Physical proximity allows for tighter feedback loops, 108

The four pillars â&#x20AC;&#x201D; 5.2

a stronger customer relationship, and the ability to promptly provide support or maintenance when needed. Therefore, we will aim at creating the first customer base in the Netherlands, drawing from her 8 Academic hospitals (two of which are in Amsterdam), 26 â&#x20AC;&#x153;top-levelâ&#x20AC;? hospitals (with all medical departments in them, dermatology included), and 3 Burn Centers. Although it may be not so relevant in the early phase, it is worth noting that the Dutch Patients association(Zorgkaartnederland (Institution) 2020) reports 89 Dermatology Centers, and that Dutch general-practitioners send patients to dermatologists with the highest referral rate of OECD countries (25 referrals per year every 1000 patients(Heijden 2013)). No market capitalization could be found specifically regarding the Netherlands. However, the global market segmentation for dermatology devices is summarized in Figure 5.3. At first, they are divided into therapeutic and diagnostic ones, with a 40:60 ratio. The estimated size of the global market for Diagnostic Devices in Dermatology varies from a lower 500 M$(Transparency Market Research n.d.) to a higher 4 bn$(Adroit Market Research n.d.), with a median of 1.5 bn$. That market is in turn segmented into: Imaging devices (our category, with roughly a 50% market share), Microscopes, and Dermatoscopes. Forecasts for the growth rate are all positive, with CAGR averaging around 5%(Adroit Market Research n.d., Transparency Market Research n.d.), the most important reasons being (1) a continued general rise in healthcare spending, coupled with (2) a general increase of disposable income and (3) increase of perceived importance of aesthetics and skin.

109

Chapter 5 — Valorization

Figure 5.3: Market segmentation for Medical Devices in dermatology. Data estimated from various sources.

5.2.3.4

Interview with domain expert

To gather information about the market, we interviewed the closest dermatology expert in our network: van Zuijlen. He is a plastic surgeon at the Beverwijk Burn Center and lecturer (bijzonder hoogleraar) at the VU University Amsterdam. On top of his work as a surgeon and lecturer, he is actively involved in the research that directly adds value to his clinical practices, and holds several peer-reviewed publications. It is worth remembering that in the past, he has already taken a “cosmetic” device and pioneered its introduction as a diagnostic tool in dermatology. Previously, he was mentioned as “co-developer” and “firstcustomer”: the overlap of his roles, and the single source for all the considerations is not optimal but we believe it suffices for the intent of this Chapter. The interview was conducted on the phone and followed the template in Chapter 5 - Appendix leaving ample room for spontaneous comments. After the phone interview, we also sent a questionnaire, kindly requesting a choice between two mutually exclusive features, and to rate the appeal of specific features. Assuming that the planned studies for feasibility and validation return positive results, van Zuijlen identified two markets that would express a strong interest in 110

The four pillars — 5.2

our device: a smaller research market and a possibly larger industrial market, i.e. cosmetics. He recognized that the latter requires a solid validation and a more mature product, simple to use and very cost-effective. Meanwhile, it is meaningful to focus the effort on appealing to curiosity-driven researchers, who are interested in original publications and pilot studies, willing to take risks and go to greater lengths to analyze and interpret the data. The first customer base could be found in the dermatology departments of the bigger hospitals; with special attention paid to the specialization of each group, and how the information provided by our device can suit their specific research needs. The combination of the suction method for mechanical characterization and the subsurface visualization, in van Zuijlen’s words, is another example of “the sum is greater than the parts”. While both techniques are independently used in dermatology research, their combination allows for the investigation of the relationship between mechanics and structure of the tissue. In particular, he mentioned that the thickness of the epidermis and the depth of the dermal junction would be the main structural parameters to correlate to the mechanical assessment. Regarding the revenue stream, the first aspects to address are the price of the device, and the decision-making process for its purchase. The most effective pricing strategy for an innovative device is not adding a markup to the labor, materials, and development costs, but rather to figure out the customers’ willingness to pay (or Consumer Perceived Value) (Hinterhuber 2008, Demirgüneş 2015). When asked about this, van Zuijlen considered (1) the price of a standalone elastometer, in the range of 20 k€, (2) that our subsurface visualization is extremely appealing, and (3) that the necessary hardware to do so is known to be expensive; he then concluded that an acceptable a price would be between 60 and 70 k€. In the analysis of the purchase decision-making process, there is a technical distinction between the end-users (in this case, dermatologists), those who takes the decision to buy (the head of the dermatology department probably has the last word) and who eventually pays for the device (the hospital has a board of directors that approves spending). However, by talking to practitioners it emerged that in Dutch hospitals and for devices below 100 k€, it is likely sufficient to secure the approval of the end-users, and that will be relayed up the decision-making chain without too much friction.

5.2.4

Value Capturing

For the first few units, it is reasonable to assume that the entrepreneurs will rely on OEM products that will be assembled at the new company’s facility via a made111

Chapter 5 — Valorization

to-order approach. This will contain the development costs and still keep a good margin while avoiding too much cashflow pressure. Therefore, the value capturing model will be based initially on direct sales to friendly customers (medical doctors who are on the front line of their field’s research activity). At this moment it is not possible to predict how this model will evolve in the future.

5.3 A Summary, on Business Model Canvases In this section, we present two business canvases, illustrating the business strategy for the short “zero-to-one sales” phase, and the subsequent “one-to-ten” phase, covering the first two years.

5.3.1

Early Phase Business Canvas: from POC to TRL 5

Figure 5.4 shows the short-term Business Model Canvas: it shows the initial strategy, on a timeframe of 6 months, to take our device from the laboratory to the market, or in other words, to realize the very first sale. Our first customer would be special, in that he could also be seen as co-developer, and wouldn’t pay the price of a fully developed product in exchange for his invaluable userfeedback to develop the device in the right direction. In fact, the instrument presented in the previous Chapter is a proof-of-concept, designed with the foresight to comply with the medical/ethical requirements for clinical trials. Yet, at this point, it is not ready to be sold to and used by a third-party: the only people able to run it are the person who built it (author of this thesis) and the medical doctor who was trained by him for the trials. At this stage, effectuation theory can be a beneficial guideline: rather than focusing on the inevitable liabilities, it’s recommendable to focus on goals that can be reached by leveraging the assets that are already at our disposition. It is a question of figuring out: what is the next step within reach that could get the business started, and how to get it done. Effectuation theory goes on and frames the process as follows: reaching that first goal (in our case, the first sale) will feed into a cycle of expanding resources and partnerships on one hand, and converging goals on the other. Consequently, in this initial phase, we are not considering the liabilities, i.e. the missing resources, capabilities or partners, but they will be discussed in the second canvas. Underlying this short-term Business model are the following basic assumptions: 1. The Medical-Ethical Technical Commission (METC) approved the application for “research medical device”

112

A Summary, on Business Model Canvases — 5.3

2. The working principle and measurements of our medical device have been validated in subsequent clinical trials (which are slated for Q32020) 3. We can rely on public grants or seed capital to cover the upfront expenses The product/market fit is the first and most important aspect to analyze. In our case, the easiest block to fill out is the “Customer Segment”: the only customer we will be aiming at in this phase is Dr. van Zuijlen. As a dermatology practitioner, his needs and wants were the main drivers of the design process. The value proposition starts from the description of what we already have: a research device to measure the mechanical response of dermal tissue under suction, while visualizing its subsurface structure. Yet, more importantly, it must answer how those features address the needs/wants from our customer: which problem(s) they solve for them, which tasks are made easier, and what will they see as added value. They are listed in the block right after the value proposition. By initially focusing on a single customer, the blocks “Customer Relationships” and “Channels” are reduced to the word direct. We can indeed count on a mature relationship of mutual trust that has evolved during the years of collaboration that led to the proof-of-concept device. The left side of the canvas is reserved for the competitive advantage: what can be leveraged to obtain it (Key Resources), how to obtain it (Key Activities) and who can help with it (Key Partners). The Key Resources have already been discussed in the Internal Analysis section. In effectual theory, those resources and capabilities help with the definition of the first attainable goals. The Key Activities reflect those assets and show what needs to be done to guarantee the realization of the value proposition. In our case, since we already secured a “co-developer” (our previous collaborator), the first step is to further develop the device, making sure it can be sold to him. To do so, in first place, a thorough testing is needed: while before the “terms and conditions” were the ones of an academic collaboration, an actual sale would involve the mutual agreement of a fully-working device. Moreover, proper documentation needs to be written, so that the device can be operated by a dermatology practitioner without a time-intensive training. It is seldom too early to discuss within the team the personal ambitions, duties, roles, equities, and maybe even an exit strategy. Premature as it may seem to list it as a Key Activity, it should not be ignored that issues that happen down the line within the team are among the top causes of startup failure (CB Insights 2019).

113

Chapter 5 — Valorization

Finally, since the device was developed with public research grants, the Intellectual Property belongs to the University. However, most modern universities appreciate the service of valorization, are enthusiastic to license the Intellectual Property, and offer Technology-Transfer Offices (at the VU it is called IXA), who help with these, and related, matters. As for the Cost Structure, we are planning to develop the first phase of the project still within the VU umbrella, i.e., without incorporating a company yet. In this way, we will be able to be hosted by the Demonstrator Lab at no cost, and still receive support from the machine shop and electronic shop of the Faculty of Science. We will therefore have only two sources of costs: the cost of goods sold (material) and 6 months of salary to sustain one of the two entrepreneurs (the author of this Chapter), who will take the lead in all the activities needed to bring the first instrument in Paul van Zuilen’s hands (including labor for goods sold). The Revenue Streams in this first phase are straightforward, as the only revenue goal is the invoice we will send to van Zuijlen. Assuming that: • • • •

the instrument will be sold for € 70,000 the CoGS-material will be € 25,000 the salary of the postdoc-entrepreneur for 6 months (CoGS-labor + part of the SG&A) will be € 25,000 other SG&A will round up to € 15,000 • there will be no investment in fixed assets and, therefore, no depreciation

during the first 6 months, the EBIT should converge to around € 5,000. Therefore, this phase can be self-sustained, as long as the entrepreneurs can find that amount of cash.

114

A Summary, on Business Model Canvases â&#x20AC;&#x201D; 5.3

Figure 5.4: Day-One Business Model Canvas: proof-of-concept to Minimum Viable Product

115

Chapter 5 — Valorization

5.3.2

Medium-Term Business-Model Canvas: reaching out

Figure 5.5 shows the Business Model Canvas for the medium term. The previous Canvas showed the strategy to get to the very first sale, which had a unique stance and “boundary” conditions. After that, we want to identify the first scaling process to reach new customers. That needs a continued effort, that start with the creation of a team and the formalization of an actual “startup”, and proceeds with the development of the Business itself, of the Hardware, and of the PostProcessing Software. Underlying the medium-term Business model are the following assumptions: 1. Our first customer will be satisfied with the device, will provide constructive feedback, and endorse the product 2. Other customers’ needs/wants will be similar enough to those of van Zuijlen’s, that they will want to buy the working device The value proposition has evolved: from a research device to a diagnostic one not a device that can be used in the clinics only as a research tool, but a fullyfledged device with all of the certifications, which can be sold as tested and validated. In fact, while the OCT imaging is an already mature technique, the mechanical characterization through suction (i.e. providing quantitative mechanical descriptors of the observed behavior) is less straightforward. If the initial value proposition was about the novel ability to visualize the displacement induced by a mechanical stimulus, the current proposition involves the refined ability to summarize that information in a few numbers (mechanical properties) with diagnostic power. The targeted customers are still not that different from our first one: dermatology practitioners that fall under the definition of innovators and early adopters: the instrument is still new and the medical field is notoriously conservative. Thus, we will aim at curiosity-driven, risk-taking dermatologists, who want to “play” with an innovative tool to get new insights, find a way to integrate it in their diagnostic procedures, and possibly publish original papers. At first, we will look for them in our geographical proximity, so that the relationship, the feedback and the device maintenance can be kept as direct as possible. After all, it is still an early phase, and the volume will not be high. Such a customer segment can be reached almost directly: initially through van Zuijlen’s extensive network.

116

A Summary, on Business Model Canvases â&#x20AC;&#x201D; 5.3

Figure 5.5: Medium-term Business Model Canvas - the first scaling and reaching out

117

Chapter 5 — Valorization

The creation of a strong team whose members have complementary skills is essential: it highly improbable for a single individual to be proficient in all the required Key Activities, which are divided in three categories: business, hardware and software development. Since we have already developed the setup as it is right now, the hardware and software aspects are much more within our own skillset than the business ones. Iannuzzi, co-founder, will support and advise on business matters, but ultimately, we will need to find someone with the appropriate skills, available time, and clear will to devote full-time to taking care of the business development for our venture. Thus, it becomes clear why Team Creation is the very first Key Activity, a conditio sine-qua-non to proceed with the business. It is hard to speculate about future Key Partners or Assets, but as Effectuation Theory suggests, the first sale (our previous goal) will definitely help with identifying the required partner. As far as the bottom part of the canvas is concerned, in the first couple of years the new company will sell instruments via direct sale. Cost structure and revenue model are therefore pretty straightforward. Following the (necessarily approximate) forecasts reported in Appendix II, we expect that the company will be able to generate an EBIT equal to 187.2 thousand € in the third year. In terms of risk assessment, it is worth recalling that, as the MedTech Innovation Report from DeLoitte (DeLoitte n.d.) puts it: “Venture capital investment in medical technology has declined over the past several years”. In fact, the development of a Medical Device is known to be a particularly risky one: the translation from proof-of-concept to marketable medical device needs to cross the so-called valleys-of-death (Páez-Avilés, Cristina, Esteve Juanola-Feliu, Islam Bogachan-Tahirbegi, Mónica Mir, Manel González-Piñero 2015, Campbell et al. 2017), common pitfalls that many MedTech companies fall into. For those firms, the “Statistics Brain” institute reports a survival rate of about 56% after 4 years(Barnhart & Peñaloza 2013). To cross the first valley of death means to successfully bring a device from the laboratory to the hospitals; in our case, the laboratory device was developed with the foresight of one day going to the clinics, so that at this point, that risk is mitigated. However, even with that first hurdle behind us, it will be crucial to secure the full support of a qualified expert for the second, and more dangerous, valley of death: the clinical translation of research devices, from clinical-research to clinical-practice.

118

Conclusion — 5.4

The development of a medical device is famously known to be a capital-intensive endeavor, and we (the founders) have no means within our affordable loss to cover for even the initial expenses. However, this liability is mitigated by several factors: access to many incubators in Amsterdam and to our University’s center for technology transfer (IXA), chance of running the expensive clinical trials with the support of the Beverwijk hospital (through van Zuijlen), and qualifications to apply for funding schemes reserved to academic technology transfer.

5.4 Conclusion As a final remark, we would like to point at the fact that within Effectuation theory, it is emphasized the importance of control over planning. The Business Canvas is an invaluable planning framework; however, physicists, exactly because of their occupation, have a strong tendency to dwell on accurate measurement of the status-quo and future planning, while forgetting that the business world is ripe with hidden variables and displays a characteristic chaotic behavior. The Lean Startup approaches this hazard by remarking the importance of tight feedback loops, shifting the paradigm from the creation of huge Gantt charts that we all know too well won’t be followed, to the iteration of its defining “build-measure-learn” cycle. As beautifully put by Sarasvathy (Sarasvathy 2001) "To the extent that we can control the future, we do not need to predict it”. Keeping this in mind, and in light of the analysis reported above, we believe that the scientific and technological results here reported have a good chance to make it to the market – a chance that we are now planning to enthusiastically take.

5.5 Disclaimer This report was written before the COVID-19 outbreak. Assumptions on CAGR, risk appetite, and both macro- and microeconomical factors may be strongly affected by the financial backlash that the pandemic crisis will produce.

119

Chapter 5 â&#x20AC;&#x201D; Valorization

5.6 Chapter 5 - Appendix 1 5.6.1

Phone Interview

Below is the guideline of the phone interview with Dr. van Zuijlen. Some answers are also found in the Chapter. Other considerations did not fit in previous sections and are reported here. Q: If the machine would work as intended, what would be its market? A: Academic Hospitals are the first choice, followed by top-level hospitals, and dermatology clinics. But in the future, do not forget the importance (and size) of the cosmetic industry. Q: Who makes the purchasing decision in the hospitals, and who would pay for the device? A: See External Analysis Section Q: How much would you be willing to pay for such an instrument? A: See External Analysis Section (approx. 60 to 70 kâ&#x201A;Ź) Q: What is the size of the hospital that could afford such price? A: Most academic and top-level hospitals can afford such device, provided they are convinced of the added value

5.6.2

Follow-up mail with form to fill in

After the interview, van Zuijlen was asked to fill in the following two tables with his own rating. This first table lists a number of choices, for which he was asked to choose the side of the trade-off spectrum he leans toward. In the second table, he was asked to simply rate the appeal of a chosen set of possible features.

120

Chapter 5 - Appendix 1 — 5.6

Table 3: Trade-offs Feature (A)

Feature (B)

Mostly (A)

Mostly (B)

Neutral

Easy to use, little training needed

Versatile, custom, requires training

30k€ with lower specs Summary of information right away

100k€, best specs available

In-depth analysis (post-processing)

Surface imaging (5x cheap)

Depth-resolved imaging (5x more expensive)

Quick, 2D imaging

Also with slower 3D imaging

Multiple measurement sub-systems Interest in cosmetics industry

Only (visco)elasticity

Interest in dermatology research

X1 [van Zuijlen] – This device is not created to be low-end technology: we should not consider it as such

Table 4: Rating features Feature Subsurface imaging Easy to sterilize Easy to carry around Silent Aesthetically beautiful Same output parameters as Cutometer See exact location of aperture during measurement

Rating [1-5] 5 This (with deformation) is the key quality 2 (for scars) - 5 (for wounds) 4 2 3 (it helps) 2 3 (preferably)

121

Chapter 5 — Valorization

5.7 Chapter 5 - Appendix 2 In this appendix, we provide a (necessarily approximate) estimate of the EBIT that the company should be able to generate in the first three years of existence. The calculation is based on the following assumptions: • • • •

•

122

Revenue is calculated as (number of units sold per year) x (€ 70,000); The evaluation of the number of units sold per year will evolve from 3 in year 1 to 5 in year 2 to 10 in year 3; The CoGS-material is calculated to go from € 25,000 per unit in year 1 to € 18,000 in year 2 and year 3 (due to optimization of the design and supply chain); The CoGS-labor is going to be moved to SG&A. This is common practice in many companies were the labor for manufacturing the goods sold is performed by personnel who spends most of their time on SG&A tasks (which include research and product development). Of course, for all research related activities, the company will ask to have access to tax benefits (WBSO); SG&A is calculated as the sum of personnel costs, rent costs, and other costs (including marketing material and activities, insurance, office material and activities, accountant). Personnel costs are calculated assuming that there will be two people in year 1 (the author of this Chapter for the technical part, and a business developer) working for the company at the minimal managerial rate of € 44,000/year (plus a fair amount of equity of the company), increasing 10% a year. Half a way in year 2, the company will hire a sale person (€ 40,000/year) and a production person (€ 30,000/year). Half a way in year 3, the company will hire another sale person and another production person. Rent costs are calculate as € 150 per sqm per year (current rate at the Amsterdam Venture Studio), according then to the space allocation: year 1 = 50 sqm; year 2 = 150 sqm; year 3 = 300 sqm. All other costs are calculated as 10% of the revenue; In the first three years, we expect to invest in computers (one per employee plus one computer for the testing lab) and basic electronic and mechanical equipment for the assembly of the devices. For the sake of simplicity, we will assume that these investments will grow with revenue at a rate of 10% of the revenue.

Chapter 5 - Appendix 2 â&#x20AC;&#x201D; 5.7

On the basis of these hypotheses, we expect the evolution of the EBIT shown in Figure 5.6.

Figure 5.6: Balance Sheet Forecast for the first 3 years. Positive amounts are in blue, and negative in red. Year 1 has the lighter tint, and Year 3 the darker.

123

6 Conclusions

The driving force that led the works presented in this thesis is the goal to efficiently and reliably measure the mechanics of skin and visualize its structure. Our first attempts employed the most accessible device we had, a micro-indenter, which â&#x20AC;&#x201C; especially at the time of use â&#x20AC;&#x201C; was found unapt to the purpose. We then analysed the reasons for that inadequacy and addressed them by improving and developing alternative methods and tools. We have investigated different approaches to the measurement of viscoelasticity, and we have integrated subsurface imaging first into a ferrule-top nanoindenters and finally into a handheld skin elastometer. The handheld device can provide the raw data for a fully-fledged OCE analysis, and its potential encouraged us to develop it further, in an effort to reach the research- and medical-devices market.

Chapter 6 — Conclusions

6.1 Viscoelasticity, virgin materials and control of sample and testing variables. Our first measurements on skin, which have not been reported in this work because of large inconsistency in the results, disregarded the time-dependent (i.e. viscoelastic) phenomena. When considering mechanical models to investigate viscoelasticity, one can distinguish two categories: time-domain, including creep, stress-relaxation and epsilon-dot methods, and frequency domain, such as Dynamic Mechanical Analysis (DMA). Each category aims at measuring the same underlying mechanical properties, but, before the work reported in this thesis, there was no experiment that compared time- and frequency- domain measurements in a systematic way, accounting for all testing and analysis variables. We have therefore carried out a focused study on this topic, which shows the compatibility of epsilon-dot (time-domain) and DMA (frequencydomain) measurements (see Chapter 2). The mechanical properties calculated through the two techniques correlate with • ≈ 0.99. Yet, the comparison showed a non-negligible offset in the data. The main source of error was likely the range of frequencies accessible by our setup, limited on the lower side by drift and experimental time, and on the higher side by the resonance frequency of the cantilever and the bandwidth of the control loop. As a result, we can directly measure only a small portion of the frequency range over which the Apparent Elastic modulus varies significantly, as shown in Figure 6.1. The high correlation of the mechanical properties obtained by the two methods is indicative that either DMA or nano-epsilon dot can reliably quantify viscoelastic phenomena; but at the same time, the offset suggests that particular care should be taken if the approaches are to be used interchangeably.

6.2 Integration of subsurface imaging into ferrule-top force transducer Skin is a layered and structured material, but the mechanical models we used shared a basic assumption of material homogeneity, which, in some conditions, may be an unacceptable simplification. Therefore, in order to make an informed decision about that assumption and eventually decide its domain of validity, we needed to visualize our sample’s internal structure, for which OCT is a perfect candidate. Furthermore, the need for subsurface visualization of the sample is ubiquitous even as a standalone technique. However, indenters usually obstruct the line of view, so that imaging is carried out either by a side viewing probe, or – provided the sample is thin enough – by inverted imaging. The cantilever in our

126

Integration of subsurface imaging into ferrule-top force transducer â&#x20AC;&#x201D; 6.2

ferrule-top is a borosilicate glass ribbon that is usually made reflective through a gold coating.

Figure 6.1: Comparison of the range of frequencies accessible through our setup to the range of significant change in storage modulus. In order to compare the mechanical properties from DMA and nano-epsilondot, we transform the frequency-dependent storage-modulus from DMA into the strain-rate-dependent apparent elastic modulus that is returned from the nanoepsilondot method. The conversion requires integration over a master curve fitting the DMA results, shown here with the blue line. The range of frequencies accessible to our device is the green area.

In Chapter 3, we have showed a ferrule-top sensor in which the cantilever is left uncoated close to its tip, which, instead of a regular microsphere, is equipped with a half ball lens â&#x20AC;&#x201C; that could act both as a spherical indenter tip, and as optical element. In this way, we could route the OCT beam through the transparent region of the cantilever and the ball-lens, integrating depth-resolved imaging into our ferrule-top nanoindenter. With that sensor, indeed, we could acquire A-scans along the indentation axis, i.e. in an epidetection scheme. We have showed how to perform subsurface imaging and indentation with the same sensor, and in the same reference frame. Such measurements, however, fall short of the full implementation of Optical Coherence Elastrography, that would have required the two techniques to run synchronously, and the collected data to be analysed accordingly.

127

Chapter 6 — Conclusions

After the publication of the article, we followed through with a first implementation of actual OCE, where we ran a loading and unloading cycle while acquiring A-scans along the indentation axis. The creation of a B-scan (2D tomogram) by transversal scanning of the OCT beam is not possible with that design, because the tip is an integral part of the OCT optics, and during an indentation stroke – in contact with sample – it can only move vertically (axially). Nevertheless, one can form a 2-dimensional image by juxtaposing those A-scans at consecutive times. The resulting image is called “M-scan”, a terminology borrowed form ultrasound imaging. We combined that with phase-resolved OCT, a well-established technique that allows the tracking of the displacement within pixels with sub-resolution accuracy. We could then process the M-scan resulting from an indentation stroke with phase-resolved OCT to track the displacements inside the sample, as shown in Figure 6.2. The development of the full implementation of OCE with our Ferrule-top multimodal sensor was suspended as we shifted our line of research onto millimetric scales with the handheld probe.

6.3 Handheld elastometer with OCT imaging: proof-ofconcept and road to market. As we realized that the millimeter scale was better suited to investigate the biomechanics of skin, we focused our effort into designing, developing, and building a clinical-research device that would incorporate all the lessons learned. The result was a device that could probe the viscoelasticity of skin while imaging its internal structure. We have shown a static modality to only consider elasticity (by simply comparing the equilibrium deformations for unloaded and loaded conditions), and a dynamic modality to include time-dependent phenomena in the analysis. We also developed an open-source software (in the form of a Python library) for the processing of the data, shown in a public repository https://github.com/LucaBartolini/OCT, that provides the basic elements for the implementation of Optical Coherence Elastography.

128

Handheld elastometer with OCT imaging: proof-of-concept and road to market. — 6.3

Figure 6.2: OCE through a ferrule top: M-Scan of the phase-differences between an A-scan and the 10th following one. We generated the figure as follows: (1) during an indentation measurement (loading and unloading), we continuously acquired A-scans along the indentation axis; (2) we extracted the phase of each pixel along the A-scan; (3) we calculated the phase difference between an A-scan and the n-th following one – in this case n=10; (4) we plotted all those differences, each as a column, to generate the image. It is possible to identify along the time-axis the contact point and the beginning of the unloading, and along the imagingaxis the cleaved end of the fiber, the “noise” created by the ball-lens, and the sample. The loading-and-unloading ramp is asymmetric, with the retraction being faster: this cause a bigger phase difference between consecutive A-scans. Image co-created with Feroldi, a co-author in (Bartolini et al. 2017).

129

Chapter 6 â&#x20AC;&#x201D; Conclusions

6.4 Vaginal Probe The initial results from the handheld device were encouraging: on that base, we devised a potential business strategy to further develop the prototype and reach the market for medical-research devices, and presented it in Chapter 4. Chapter 5 discusses the road to market for the probe as a tool for dermatologists; however, we later learned about the possibility of translating the underlying technology to enable transvaginal measurements. In particular, we have met with gynaecology practitioners, who identified a compelling need for the mechanical characterization and subsurface visualization of the vaginal wall. Subsurface structure and mechanical behavior, indeed, are strongly linked to pelvic floor dysfunctions, one of the most common reasons for gynaecological surgeries. However, unlike in dermatology, there is currently no offer for a medical device fit for that purpose. Therefore, we have decided to aim for this niche, further developing our probe, and changing it form factor â&#x20AC;&#x201C; schematically shown in Figure 6.3, so that it can be used for transvaginal elastography measurements.

Figure 6.3: Concept design of a transvaginal handheld OCE probe, that integrates OCT and suction elastometry.

Considering all of the above, our future efforts will be driven by a strive for valorization and clinical translation, to reach those practitioners that could benefit from a powerful, new diagnostic tool for the visualization and mechanical characterization of epithelial tissues.

130

7 Appendix 1 Excerpts from the “Investigative Medical Device Dossier”

The Investigational Medical Device Dossier (IMDD) contains all the information needed for the Medical-Ethical commission to allow a research-device to be used in a clinical study. As the Central Committee on Research Involving Human Subjects, states: “The IMDD specifies all items that must be covered for the application to the review committee (MREC or CCMO) for non-CE-marked medical devices intended for clinical investigation.”(CCMO 2020). At the time of this writing, the submission has not yet been completed. Below, we report the considerations about classification and risk-management. They are especially relevant sections of the IMDD because they define and estimate possible risks for the patient or the operator, and discuss the implemented solutions to mitigate them.

Selected excerpts from the IMDD document that is being compiled in collaboration with Ludo van Haasterecht and Jelmer Weda.

Chapter 7 — Appendix

7.1 Classification Reference document: Regulation (EU) 2017/745, ANNEX VIII (EU Parliament 2017)

7.1.1

Applicable definitions:

1.1. “Transient” means intended for continuous use for less than 60 minutes. 2.5. ‘Active device intended for diagnosis and monitoring’ means any active device used, whether alone or in combination with other devices, to supply information for detecting, diagnosing, monitoring or treating physiological conditions, states of health, illnesses or congenital deformities. 2.8. Injured skin or mucous membrane’ means an area of skin or a mucous membrane presenting a pathological change or change following disease or a wound.

7.1.2

Applicable classification rules

7.1.2.1

Rule 4

All non-invasive devices which come into contact with injured skin or mucous membrane are classified as: • •

• •

class I if they are intended to be used as a mechanical barrier, for compression or for absorption of exudates; class IIb if they are intended to be used principally for injuries to skin which have breached the dermis or mucous membrane and can only heal by secondary intent; class IIa if they are principally intended to manage the microenvironment of injured skin or mucous membrane; and class IIa in all other cases.

This rule applies also to the invasive devices that come into contact with injured mucous membrane. 7.1.2.2

Rule 10

Active devices intended for diagnosis and monitoring are classified as class IIa: • if they are intended to supply energy which will be absorbed by the human body, except for devices intended to illuminate the patient's body, in the visible spectrum, in which case they are classified as class I; 132

• if they are intended to image in vivo distribution of radiopharmaceuticals; • if they are intended to allow direct diagnosis or monitoring of vital physiological processes, unless they are specifically intended for monitoring of vital physiological parameters and the nature of variations of those parameters is such that it could result in immediate danger to the patient, for instance variations in cardiac performance, respiration, activity of the central nervous system, or they are intended for diagnosis in clinical situations where the patient is in immediate danger, in which cases they are classified as class IIb. Active devices intended to emit ionizing radiation and intended for diagnostic or therapeutic radiology, including interventional radiology devices and devices which control or monitor such devices, or which directly influence their performance, are classified as class IIb.

7.1.2.3 Rule 11 Software intended to provide information which is used to take decisions with diagnosis or therapeutic purposes is classified as class IIa, except if such decisions have an impact that may cause: • •

death or an irreversible deterioration of a person's state of health, in which case it is in class III; a serious deterioration of a person's state of health or a surgical intervention, in which case it is classified as class IIb. Software intended to monitor physiological processes is classified as class IIa, except if it is intended for monitoring of vital physiological parameters, where the nature of variations of those parameters is such that it could result in immediate danger to the patient, in which case it is classified as class IIb. All other software is classified as class I.

7.1.3

Classification result:

Classification: IIa, given: contact non-injured skin, active device for diagnosis, non-visible light emission, software for clinical evaluation.

133

Chapter 7 â&#x20AC;&#x201D; Appendix

7.2 Risk Management Plan Reference document: ISO 14971:2009

7.2.1

Scope

The device is intended for clinical evaluation. The scope of the study is to evaluate the effectiveness and performance of the proposed method. Currently, the device is under development. The risk management process will be adapted and updated on the hand of the progress on the development of the device. Risk management must at least be reviewed prior to measurements on volunteers participating in the clinical evaluation.

7.2.2

Requirements for review of risk management activities

Risk management activities should be reviewed after implementation of risk control measure. Risks should be re-evaluated in the case of adverse events, a changed intended use or any other reason to change the system's design.

7.2.3

Criteria for acceptance of risk acceptability

Since the device is developed to evaluate its effectiveness and performance and it is used on human volunteers, the use of the device should not lead to any unacceptable risks.

7.2.4

Verification activities

Effectiveness of optical, electrical and pressure risk control measures must be verified prior to any measurement on human subjects, and after the first clinical evaluation.

7.2.5

Method for obtaining relevant post-production information

When necessary, the risk management process will be evaluated after user feedback, feedback from the subject population or accidents.

7.3 Risk analysis Intended use of the system defined as â&#x20AC;&#x153;Dermal tissue elastometer with 3D imagingâ&#x20AC;?: This device is manufactured by the VU University Amsterdam, shall be used exclusively for clinical evaluation and is only to be operated by qualified 134

personnel. The device is for research purposes only and no diagnosis should be made based on the results of the measurements. See chapter 2 for a detailed description of the system

7.3.1

Identification of hazards and risk estimation:

Main hazards are optical radiation, electrical currents and too-negative pressure. Other accountable hazards include software and user errors, biological hazards, and mechanical hazards. The following section describes risks estimation without any risk control measures. 7.3.1.1

Optical hazards

The source of the device emits optical radiation in the infrared portion of the spectrum: its power, deliverable dose and its interaction with the software should be assessed. 7.3.1.2

Electrical hazards

The energy to the device is supplied by a 230 V mains connection. Electrically non-conductive parts are supposed to be in touch with the patient and the operator; electrically conductive parts may come in contact with the patient and the operator. Although the device is used non-invasively, there are electrical hazards present. 7.3.1.3

Pressure hazards

The elastometer accessory applies a mechanical stimulus to dermal tissues by applying a negative pressure. If the pressure is too low (i.e. negative pressure too high), the patient might experience discomfort or tissues might be damaged. 7.3.1.4

Biological hazards

Multiple subjects are measured with the device. The elastometer chamber (front accessory), and by proximity, the handheld probe, will come into physical contact with the skin of different people. This leads to the possibility of transmission of infectious diseases. 7.3.1.5

Mechanical hazards

The device consists of precision optomechanical components and electronics placed on a movable medical cart (see appendix 6). The total weight is estimated to be around 100 kg. The handheld probe is mounted on a cable of around 3m, that puts distance between the cart, and the operator and patient. 135

Chapter 7 â&#x20AC;&#x201D; Appendix

7.3.1.6

Software hazards

The device is operated by software, which is custom developed and subject to change and/or failure. Moreover, the software returns the OCT images and the measurable quantities (skin elasticity and optical properties), which might be wrong. 7.3.1.7

Thermal hazards

The scanning operation of the galvanometric mirrors produces heat.

7.4 Risk evaluation 7.4.1.1

Optical hazards

The light source is a class 1M laser under IEC 60825-1:2014, which means that it is safe under all conditions of normal use. This classification means that the maximum permissible exposure cannot be exceeded, even when fixating the laser with the naked eye (the tissue with most restrictive safety thresholds for light exposure). 7.4.1.2

Electrical hazards

Hazardous situations from electric shocks are present but remote. The system's energy is supplied by 230 V mains which is at a dangerous level. Moreover, lower voltage open contacts may be present on the connection between the handheld probe and the scanning controller. Risk control measures for electrical hazards should be taken. 7.4.1.3

Pressure hazards

A miniature diaphragm pump provides the negative pressure required to apply a mechanical stimulus. The pump has a rated Ultimate vacuum of 100 mbar. This pressure might result uncomfortable if applied on hypersensitive or injured tissue, but it is unlikely to damage the tissue. Risk control measures and for pressure hazard should be taken. 7.4.1.4

Biological hazards

The risk of cross-infection is likely since multiple subjects are in physical contact with the elastometer accessory chamber, and by proximity, with the handheld probe. Risk control measures for cross contaminations are required.

136

7.4.1.5

Mechanical hazards

The setup consists of heavy equipment and is installed on a movable medical cart. During measurements, it is not necessary to move the cart. The measurements are taken by the handheld probe, which is connected to the setup by a 3m long cable. Mechanical hazards are estimated as remote. 7.4.1.6

Software hazards

Fault conditions from software are likely to arise from user errors or software instability. The software is used to control the applied negative pressure, to prescribe the movement of the scanning mirrors and process the images: any of these operations is subject to software failure, which in turn may lead to considerable levels of optical radiation or too-high negative pressure applied to dermal tissues. 7.4.1.7

Thermal hazards

The motors that move the mirrors are fixed on a metal mount, in turn mounted on a metal plate. At temperatures above 238 Â°C, POMC releases formaldehyde, which is a toxic and irritant gas.

7.5 Risk control 7.5.1.1

Optical

The optical radiation source is classified as Class 1M laser, and as such, is inherently safe under all conditions of normal use. Therefore, no actual risk is posed by the laser source and no additional risk prevention methods are needed. 7.5.1.2

Electrical

The Electrical risk control measures are schematically illustrated in appendix 11. General electronic equipment standards are used to reduce risks concerning electric shocks (NEN-3140 & NEN-EN 50110-1:2013). An IEC-60601-1 certified isolation transformer (appendix 4) is also implemented, to reduce the risk of electric shock resulting from a connection from subject or user to the ground. All accessories of the device should be connected to this isolation transformer. The conductive parts accessible to the operator and the patient must be connected to the additional protective earth (PE) terminal. Moreover, the frame of the medical cart on which the setup is mounted is all-conductive and will be grounded to ensure an additional level of electrical safety, conforming to the IEC-60601-1 standard. The handheld probe, which is the only accessory coming in contact with the patient is designed conforming to the requirements of IEC-60601-1. The 137

Chapter 7 — Appendix

device should be tested and retested conforming to the requirements of a class 1 electrical device, with BF-type applied parts, and their corresponding Means of Protection. Moreover, the frame of the medical cart on which the setup is mounted is all-conductive and will be grounded to ensure an additional level of electrical safety. 7.5.1.3

Pressure

By design, the suction line is not intended to hold sustained vacuum: the minimum pressure reachable in the suction chamber is higher (i.e. suction is lower) than the ultimate pressure of 100 mbar declared in the pump specification. The minimum pressure reachable in a sealed suction chamber was measured to be 250mbar (absolute), i.e. 750mbar suction (with respect to the atmospheric pressure), after 300 s (five times the longest intended experiment). In a clinical setting, while measuring on in-vivo dermal tissue, the minimum reachable pressure will be even higher, due to the additional fact that skin does not seal the suction chamber, and that since pressure decreases asymptotically, a shorter experimental time translates to a higher absolute pressure. 7.5.1.4

Software

Malfunctions or operator errors in software cannot result in radiation damage to dermal tissues (the laser source is Class I laser) or in damage to the skin (the ultimate pressure of the pump is inherently safe). 7.5.1.5

Biological

The risk of cross-infection between subjects is reduced by disinfection of the chamber accessory with 70% ethanol after each measurement session. Please see Ethanol Safety Data Sheet (MSDS) for further information. 7.5.1.6

Mechanical

The entire setup is installed on a medical cart conform to the mechanical requirements of IEC-60601-1. All the equipment is fixed with screws on the cart, or if not possible, placed on anti-slip mats. The cart has breaks on its wheels, so that it can be put in place. Measurements are taken far from the cart, with the handheld probe extending up to the length of its cable (3m). 7.5.1.7

Thermal hazards

The handheld probe is equipped with a Temperature sensor that turns off the controller in case of overheating. The Temperature is set to stay below 55 °C, well below the threshold for release of formaldehyde by POMC (238 °C). 138

7.6 Implementation of risk control measures Since the device is currently under development it is not yet possible to perform measurements with the device. However, the following risk control measures should be implemented while developing the device: 7.6.1.1

Electrical

Already in the development stage, all the electrical components are connected to the isolation transformer. The frame of the medical cart, and all electrical appliances are connected to protective earth. These safety measures already provide a higher level of protection to the user with respect to the basic safety standards (NEN-3140 & NEN-EN 50110-1:2013). In-house electronics (such as the galvanometric mirror driver) should be developed and tested as class 1 electrical devices and BF-type (basic insulation) applied parts. 7.6.1.2

Pressure

The software performs a check to make sure that the pressure regulator is never set to an absolute pressure lower than 500 mbar. Furthermore, we have mounted a venting valve along the vacuum line, 1m away from the handheld probe. The operator can open that valve to quickly vent the whole line. 7.6.1.3

Biological

Cleaning the elastometer chamber and the whole handheld probe with ethanol should be a daily routine part of operating the device. This is documented in the manual. 7.6.1.4

Mechanical

Mechanical risk control measures are implemented in the design of the cart. 7.6.1.5

Thermal hazards

The electronics that reads the Temperature and switches the controller off is always running during operations.

7.7 Residual risk evaluation 7.7.1.1

Pressure

A literature study reveals that the most in-vivo suction tests reach a suction of 500 mbar(Pierard et al. 1995, Draaijers et al. 2004), a level considered safe in all cases. One study, reported an in-vivo suction of 600 mbar(Pedersen & Jemec 139

Chapter 7 â&#x20AC;&#x201D; Appendix

2006) where an increase of transepidermal water loss is measured, but no mechanical damage is found. In case our software would fail and the safety valve be left close, a suction higher than 600mbar and up to 750 mbar (as from our leaktest) could be applied, a risk that we estimate to be low. A maximum suction of 750 mbar, could be applied the following events would realize one after the other: (1) the software control fails, (2) the safety valve is left closed, (3) the probe is kept in contact with the patientâ&#x20AC;&#x2122;s skin for long enough time (more than 30seconds). Considering that the Optical radiation source is inherently safe, and that electrical hazards are appropriately taken care of, all residual risks are acceptable.

7.8 Risk-benefit analysis Since the device is exclusively for clinical evaluation there is no benefit for the subject (or patient) yet. Therefore, the use of the system should not lead to unacceptable risks. When risk control measures are correctly implemented and verified, the risks of using this system are acceptable with respect to the intended use.

7.9 Risks arising from risk control measures Correct implantations of risk control measures are assumed, when correctly implemented following risks are arising from risk control measures: 7.9.1.1

Biological

Ethanol is flammable, precautions should be taken when the disinfection procedure is followed.

7.10 Completeness of risk control Currently, the elastometer device is under development, foreseen risks are described and evaluated. Risk control measures are proposed, evaluated and should be implemented. In all probability, new potential hazards will be found while further developing the device and this should be evaluated at a later stage. Suggested reassessment interval for reviewing risk management process is 1-2 times a year.

140

7.11 Evaluation of overall residual risk acceptability The sequence of events that would lead to the ultimate suction of 750mbar to be applied to the skin is highly unlikely. Moreover, in the absence of reported literature on the topic, one can reasonably assume that the pressure differential required to rupture skin through suction (i.e. by locally applying a pressure lower that the atmospheric one) is comparable to the differential required to rupture skin through positive pressure (i.e. by locally applying a pressure higher than the atmospheric one) or tension. The latter, also called tensile strength of skin, has been reported to be in the range of 5 to 30 N/mm2 (Vogel 1987, Edwards & Marks 1995), which – even at its lower boundary – means a pressure of 50 million mbar (i.e. 50 000 atmospheres) to rupture skin. The – unlikely – application of 750mbar of suction is far from any rupture damage. In the scope of the intended use of the elastometer device, the presence of unacceptable risks is not appropriate. After risk control measures have been implemented, the level of safety is in accordance with standards and common practice (ethanol disinfection) and does not lead to any unacceptable risk.

141

Chapter 7 â&#x20AC;&#x201D; Appendix

Figure 7.7.1: Failure modes and Effects Analysis (FMEA)

142

8 Summary

Scientific evidence and human intuition agree about the close relationship between a tissueâ&#x20AC;&#x2122;s mechanical behavior and its condition. Ordinary examples are the changing elasticity of a wound as it heals, the increasing brittleness of bones as they age, and the softening of a fruit that ripens. Thanks to their predictive power, mechanical properties are used as a patho-physiological indicator in medical disciplines, for the diagnosis of diseases, evaluation of therapies, surgical and prosthetic intervention. The quantitative methods for mechanical characterization, however, were developed in the last two centuries to investigate inorganic matter such as metals, ceramics and polymers. We can identify an ongoing translational effort to allow those methods to be applied for the objective quantification of the mechanical properties of all types of soft biological matter. This dissertation reports our endeavor to measure the mechanical properties of dermal tissues through tailored approaches; first using established tools and then developing our own. We will discuss how and why our initial approach â&#x20AC;&#x201C; microindentation â&#x20AC;&#x201C; fell short of the purpose yet imparted an impulse to the following research. We found an opportunity in that situation by taking a closer look at the shortcomings of the instruments we used, to finally develop the original solutions that constitute the backbone of this thesis. The capstone project, involved a multidisciplinary effort to design, build, and run preclinical tests with a medical device to both assess the elasticity of skin and obtain its subsurface 3D structure. In Chapter 1, we introduce skin anatomy and explain why its mechanics is both interesting and relevant for this study. Moreover, we illustrate the basic principles of nanoindentation and the novel technology (ferrule-top) that implemented it. We then analyze the reasons for which the nanoindentation approach was unsatisfactory for our purpose, and how the review of those limitations drove our research forward. Before the chapter ends, we discuss the fundamentals of a depth-resolved imaging technique called Optical Coherence Tomography, which will be integrated with mechanical characterization in the following chapters.

Summary

In Chapter 2, enabled by the novel ferrule-top technology at hand, we validate a model to directly compare viscoelasticity measurements conducted in the timedomain to those conducted in the frequency-domain, using our unique ability to measure in both domains while keeping identical sensor and sample conditions. In Chapter 3 we present a new design of a multimodal ferrule-top sensor that integrates depth-resolved imaging in our microindenter geometry, so that a multimodal characterization – optical and mechanical – can be performed by a single integrated sensor. The work demonstrates a proof-of-concept setup in which the two acquisition modalities are not concurrent. Nevertheless, it represents the first step toward a more intimate combination of the two techniques – i.e. elastography – that would allow to produce spatially resolved maps of the mechanical properties beneath the indenter. The works in Chapters 2 and 3 result from our intention to address the shortcomings of nano- and microindentation: the need for a reliable approach for measuring time-dependent behavior, and the issues arising when measuring a non-homogeneous (in our case, layered) material. We eventually concluded that in order to have an instrument that could impact clinical practice, we would need to reach a much bigger scale and sense the mechanics of dermal tissues as a whole, meaning we would need to affect and measure our sample at the scale of millimeters rather than micrometers. Chapter 4 shows the clinical-research device that we developed and built to investigate the mechanical properties of skin at that scale, and that includes subsurface imaging via Optical Coherence Tomography. The instrument measures the mechanical properties no longer with indentation, but with a suction technique already established in dermatology. Dermal tissue is drawn by a small negative pressure (suction) through a circular aperture in a suction chamber. Skin’s elasticity determines the amount of compliance into the aperture, while our imaging technique provides depth-resolved imaging of the interested region. The device can characterize the mechanical response, image skin’s substructure and holds potential for elastography. Pending regulatory approval, we plan to use it in clinical trials. Chapter 5 lays out a potential business strategy for the clinical translation and commercialization of the research device presented in the previous chapter. The discussion is articulated around the Business Model Canvas, a tool which provides a systematic overview of the business considerations that are necessary in undertaking such an endeavor.

144

9 Acknowledgements

I must admit that writing these acknowledgements was tough: I was overwhelmed with gratitude, and – classic me – I postponed the job for a while. Of course, I ended up writing the following in the last days before sending everything to the printshop, but here are all my thank you. There is an age when a kid grows out of the absolute conviction that his parents are superheroes. He just starts seeing them as humans, with all their shortcoming and their mistakes. That never happened to me (someone might say it is because of my peter-pan syndrome) but I am convinced you – Mamma&Babbo – are real superheroes. I’d be proud of myself if I could ever find a fraction of your strength, dedication, and unconditional love. Thank you for your relentless support! Davide, you have been enormously inspiring over these years. You’ve been down-to-earth and layed the basis for an authentic and solid relationship, one in which we could fully acknowledge our divergences, which is especially remarkable for someone with your outstanding achievements. It’s a privilege to still be around you. Thanks for what you taught me and the support you still provide. Paranymphs, colleagues and friends, everyone who brought joy, chance for growth, awe, and inspiration over these years: Marica, I don’t even know where to begin! The course of our PhDs was deeply intertwined, and you were among the first I would turn to for advice for work and life matters alike, and it was always thoughtful. I’m also thankful for your relentless push to stick together: many nights I would have groggily left, when your contagious energy brought a second wind. You are deeply missed! Ludo. It happens to meet people that become less and less interesting as one hears them repeating old stories and stale jokes. Well, you’re exactly the opposite, the more I know you, the more of a great person you become to my eyes. Back when I was developing the medical device, you embodied the end-user that I wanted to make happy: the outcome of that project would have been way worse without your early feedback and interest.

Acknowledgements

Kingson for inspiring me with your hard work and brilliant takes; for bringing diversity to our group, I fondly remember our conversations during lunch and the many times your perspective was new, lateral, and great food for thought. Nelda for the fun (the others had) when we were limping around in San Francisco, for helping with the SPIE chapter and the countless breaks with Marica Martin Slaman, you coffee break partner, stubborn relentless helper. Your teachings were invaluable, I knew I could always count on you for a quick answer, or a substantial help, even in your busier times. Our coffee break stronghold was often a battleground for ideas, and the emails we’d share afterwards could be a wonderful collection of trivia facts Ata, crazy officemate, I learned a great deal from you, from physics to geopolitics, and no less about life in general. You literally and figuratively left a void when you flew first-class to the US and I still miss our copious exchange of “bullshit cards”. Thanks for having stuck around. Leo, opened so many windows about the Dutch culture. You and your training made me race my first trail run and first half-marathon. We shared climbing sessions, cycling and nerd nights discussing hydrogen power and photosynthetic LED! I’m looking forward to many more Hedde you were – and still are – a role model, with a wisdom that goes far beyond the boundaries of science, and with the skills to make the backbone of Labview code for all nanoindenters! Your goodbye left a hole in the group, and I wished many times you could have been there for some valuable advice René, your goodbye presents – the Stroh 80° and the 24h card – are to this day some of my favorite! Thanks for showing me the cherry blossom park, the bread making way and having us for those unforgettable summer solstice drinks! People from the section, even if the overlap of our project was less that optimal, I had a great time hanging out with all of you in so many occasions! Even just corridor chats are dearly missed in these corona times! Imran for you disarming kindness, your words, tips, and guidance have been extremely valuable! Erik for the movie and series suggestions, and the good lunchtime chats. Steven and Camiel for the precious advice of the early days at the VU and the good company. Joost for the fun times getting ready running in the darkest corners of Sloterpark for the one of the first events that would end up being canceled for the corona! Giorgio Mattei for the pain and fun of writing a paper together for about 18 times. Massimiliano, you are among the few people whose preparation I’d define “intimidating”, yet you carry it with an incredible lightness. Jelmer, for sharing with me so much of your knowledge, from the technical file, to how to deal with 146

Acknowledgements

the workshops, to the soft lithography techniques. Agata, for the many corridor jokes, and for helping me upcycle my office material. Maggie for the fun in all our (countless) SPIE events. Thanks Kari, Mathi, Andy, both Max’s, Denise, Maggie M. The parties in Veldhoven, for the great company at Marica’s wedding, for the Tuinzaal Friday-drinks when we’d call the Security desk to have the door on the garden opened! Max, you are like a brother, for our (bike) trips, cooking sessions and singing ones! Our 6-wheeled Canadian adventure and all the two-wheeled Dutch ones. Your presence in the moment, your way with words and your wildcard unpredictability are something to look up to! Thanks also for proofing parts of this book, and for your precious writing and editing tips: especially the unforgettable “make a mess then clean it up” Sevgi you have been a pillar! Energetic, talkative, amazing friend. You’ve been there for each of my steps, cheering for me, always there for a silly or a deep talk, a walk in the park, a game of beach volley or a dinner party. To many more! Andrea and MC, it is heartwarming to know we can always get together and chat away life’s troubles, eat whatever as long as it is exaggerated, or play a game of squash. Of course, with the background constant of plotting the next unicorn business. Everything I’d write would be reductive. Fabio, I will be forever grateful for many things, including when back in 2014 you brought to my attention the PhD vacancy that eventually lead me here. A great deal of my social life and work output can be traced back to some spark you generated. Having you as flat-mate for years was intense: fun, tough and inspiring at the same time, there was always something new to learn from you. I’m thankful for all of that. Chiara for making this beautiful cover, bringing some sunshine in my second amsterdamse winter and, most importantly, for introducing me to your magnificent husband Andrea. I have the fondest memories of your time in Postjesweg and I am really glad you are still around! Cocca, who felt like an old friend from day three of knowing you. Your heartwarming presence made Amsterdam feel more like home. Maien – Schamaien – we joked that I’d mention the fun story of draining the septic tank of the RV rented few hours before, with none of us having ever done that, in the middle of the night. Here it is, it’s public now. You’re a great friend who’s never afraid to bring up the tough conversations. Keep it up!

147

Acknowledgements

Marione, we were an inseparable couple in SVU and we had so much fun, we started climbing together! Even after so long from our daily interaction, it still feels like I could drop by at any moment and have a three-hour chat with you. Michele Monti, Jacopo Solari and the expanded Levantkade crew (with a special mention to Tarq e Titta) every time I met with you guys it felt like a giant family coming together. I felt so lucky and in awe when I got to come to your cinema nights and dinners and nights out, good times! Douwe for being a real friend, I love our healthy “competition” and how we can create an argument about the most disparate things, it’s always stimulating and a great chance to learn new things. Thanks for those growth opportunities, and also for all the parties at your place and the fun at our beach volley games. Fede, normal people must pick two between social life, work achievements and sleep. But it seems you nail them all, and I keep look up to you, trying to learn something. Plus, your giant heart shows in the tight knit circle of friends you have around! I’m proud of being counted among those! To the Ardennes climbing crew, Felix, Jas, Diego, Francesca who kickstarted one of my dearest passion! Alberto Natali “Maestro” you motivated me and showed me the way up in so many climbing routes. It was always great fun to squeeze our bodies to the limits next to someone like you! Radek, you were a great random encounter, thanks for our crazy climbs – above all the Zugspitze – and for the Mate cup that I still use these days! Best luck in your new german life, hope we’ll meet again soon! Gioele, not only you are one of my best friends, inspiring me with your perseverance, endless skills, and stark deliberate choices, you are also the one with whom I shared the most physical pain. I will never thank you enough for jumping in on the crazy idea for the “dissertation completed” celebration-hike Bologna to Montiano. Can’t wait for the next one! Splinter: you’ve had a much bigger influence in my life than I like to admit! Including the passion for Budo which led me to Ulm and your gentle pushing me towards Cristina on that infamous night – some of the biggest influences of my life, and which I am extremely grateful for! I’d say…keep it up Nicola, I like to say that when in Germany, I wanted to stay away from other Italians, but couldn’t stay away from you. You are the great synthesis, part “Italiano medio” with all the bad jokes, and part one of the most brilliant and deep-thinking people I know! Thanks for the holidays, our scary first via ferrata and all our enlightening chats! 148

Acknowledgements

Alessio, giant friend who stuck through high school, university, and PhD: one day we’ll have our silly empire. For the time being, we’ll find happiness in a dinner with some crostini in Saiano, good wine and plans of mycelium empires. Valentina Mazzoli because you are inspiration for work ethics and social living, I think I learned something from you about the ability to make things work and be totally into what I’m doing Cole, when you’re around it feels like home (maybe because we basically shared one for so long!) Thanks for all your visits in Amsterdam, and our countless conversations. I have the impression that without your direction, I am sort of following your steps, with a few years delay. Does that mean that in a few years I’ll fix a freakin’ vineyard? No clue! Vova above everything I am thankful because you introduced me to Yarik the great. But well, also because you lifted the veil of Whiskey appreciation, you are one of the best sparring partners for ideas and you share my absolute love for dinosaurs. Oz, meeting you up again was a prodigal son moment! Thanks for sharing and helping me process the toughest times from the PhD, and the good times in Utrecht. Also, I use your slides trick every day and it is making a huge difference in my work life! Thank you so much. Nick, our backcountry hiking gave me some of the best memories in my life and that say it all. Longing for the next one! The AMOLF crew, Tzeni (who also made San Diego a great experience), Giorgio (who gifted us the memorable “you can always bring it back into the trash pile”), Agata, Parisa, Mario, Giulia Giubertoni. Sofia, you were a pillar presence in Cesena, one that I never skipped seeing when I am back, and not simply because we’re as close to neighbor as it gets in the countryside. Thanks for your philosophy lessons and stories, our walks, and for bringing me along to all the great activities! All friends from back where I grew up: Massi, Ele, Edgar, Andó, Cate, Paves, Cugi, Frez, Biga, Checca, Cate, Corza, Giuli, and the ladies Jenny, Valina and Vale – for the hikes, the dinners, the outings at the river and at the Sand Gate. You all make me look forward to being back in Romagna and make me miss it when I’m away! SVU heren 3 – for all the bad trainings, memorable games, rugged tournaments, terrible chants, and wild parties! We may have been the underdogs of 3rd class, but the camaraderie and the fun was real. Special thanks to Robin, Budgie, Kevin, Flote, Sjon, Sebas and Giel, plus of course our explosive trainers Elza and Kees, 149

Acknowledgements

and the rest of the SVU club! You guys might have a speck of responsibility for my delay in finishing the PhD, but I miss you all nonetheless. 30 MHz team: Flavia and Jasper for giving me the chance, Fokko for teaching me so much about coding, Rinke for the laughter, Helena for the comradery, Daan for the great tips and Niels for teaching me a few lessons in pingpong. But above all, a huge thanks to the whole team for making me feel included from the very beginning, even though I was in the PhDâ&#x20AC;&#x2122;s desperation valley. I felt part of a family, useful and with a purpose while working immersed in the green sea of leaves inside the Pantar greenhouse! It was an amazing experience. A final extended thanks for having taken a few steps beside me in these years to Rector, Michael and Lucy, Serena Gandolfi, Ariana, Maaike, Valentina Davidoiu, Vlad, Vincent Friebe, Nicoletta, Juri, Sara, Luca Bersanini, Vincenzo, Luca Ferrari, Ravi, Gus, Daniel Foster and Danny, Ceren, Ale, Eliza, Claire, Davide Taviani, Dorotea, and finally to anyone I might have forgotten. Crispies, it is hard to imagine how I would have accomplished this without you. I could always count on material help, but most importantly you gave me unconditional love and support. You were the calm within the storm (and sometimes the storm too) â&#x20AC;&#x201C; I would not change a thing. Grazie amore mio.

150

10 References Adroit Market Research. (n.d.) Dermatology Devices Market Size – Global Industry Type, Trends & Forecast 2025. Retrieved March 2, 2020, from https://www.adroitmarketresearch.com/industry-reports/dermatologydevices-market Afdhal, N.H. (2012, September) Fibroscan (transient elastography) for the measurement of liver fibrosis. Gastroenterol. Hepatol., Millenium Medical Publishing. Agache, P.G. & Humbert, P. (2004) Measuring the Skin, 49, 784. Retrieved http://books.google.co.uk/books/about/Measuring_the_Skin.html?id=APiLX8aBakkC&pgis=1

from

Alexander Osterwalder, Y.P. (2010) Business Model Generation: A Handbook for Visionaries, Game Changers, and Challengers, John Wildey & Sons. Andreesen, M. (2007) EE204 Business Management for Engineers and Computer Scientists - Product/Market Fit. Stanford Univ. . Retrieved March 4, 2020, from https://web.stanford.edu/class/ee204/ProductMarketFit.html Arnaud, B. (2018) The 10 most common reasons why startups fail. EU-startups. Retrieved March 5, 2020, from https://www.eu-startups.com/2018/09/the-10-most-common-reasons-why-startups-fail/ Attendu, X. & Ruis, R.M. (2019) Simple and robust calibration procedure for k-linearization and dispersion compensation in optical coherence tomography. J. Biomed. Opt., 24, 1, SPIE-Intl Soc Optical Eng. doi:10.1117/1.jbo.24.5.056001 Baniasadi, M., Xu, Z., Gandee, L., Du, Y., Lu, H., Zimmern, P. & Minary-Jolandan, M. (2015) Nanoindentation of Pseudomonas aeruginosa bacterial biofilm using atomic force microscopy. Mater. Res. Express, 1, 045411. doi:10.1088/2053-1591/1/4/045411 Barney, J. (1991) Firm Resources and Sustained Competitive Advantage. J. Manage., 17, 99–120. doi:10.1177/014920639101700108 Barnhart, M. & Peñaloza, L. (2013) Startup Business Failure Rate By Industry | Statistic Brain. J. Consum. Res. doi:10.1086/668536 Bartels, V.T. (2011) Contact sensations of medical textiles on the skin. in Handbook of Medical Textiles, pp. 221– 247, Elsevier. doi:10.1533/9780857093691.2.221 Bartolini, L., Feroldi, F., Weda, J.J.A., Slaman, M., Boer, J.F. de & Iannuzzi, D. (2017) Multimodal probe for optical coherence tomography epidetection and micron-scale indentation. J. Innov. Opt. Health Sci., 10, 1742007. doi:10.1142/S179354581742007X Bednár, I.R. & Tarišková, I.N. (2018) Indicators of startup failure. Int. Sci. J. ‘Industry 4.0’, Vol. 5, Scientific Technical Union of Mechanical Engineering ‘Industry 4.0’. Retrieved from http://www.eban.org/about-angelinvestment/early-stage-investingBeekmans, S. V. & Iannuzzi, D. (2015) A metrological approach for the calibration of force transducers with interferometric readout. Surf. Topogr. Metrol. Prop., 3, 025004. doi:10.1088/2051-672X/3/2/025004 Beekmans, S. V. & Iannuzzi, D. (2016) Characterizing tissue stiffness at the tip of a rigid needle using an optomechanical force sensor. Biomed. Microdevices, 18, 1–8, Springer US. doi:10.1007/s10544-016-0039-1 Beekmans, Steven V., Emanuel, K.S., Smit, T.H. & Iannuzzi, D. (2017) Minimally Invasive Micro-Indentation: Mapping tissue mechanics at the tip of an 18G needle. Sci. Rep., 7, 11364. doi:10.1038/s41598-017-10526-4 Benítez, R., Moreno-Flores, S., Bolós, V.J. & Toca-Herrera, J.L. (2013) A new automatic contact point detection algorithm for AFM force curves. Retrieved from www.r-project.org Berardesca, E., Maibach, H.I. & Wilhelm, K.P. (2014) Non invasive diagnostic techniques in clinical dermatology. (E. Berardesca, H. I. Maibach & K.-P. Wilhelm, Eds.)Non Invasive Diagnostic Tech. Clin. Dermatology, Berlin, Heidelberg: Springer Berlin Heidelberg. doi:10.1007/978-3-642-32109-2 Berends, H., Jelinek, M., Reymen, I. & Stultiëns, R. (2014) Product innovation processes in small firms:

References Combining entrepreneurial effectuation and managerial causation. J. Prod. Innov. Manag., 31, 616–635, Blackwell Publishing Ltd. doi:10.1111/jpim.12117 Bertassoni, L.E. & Swain, M. V. (2014) The contribution of proteoglycans to the mechanical behavior of mineralized tissues. J. Mech. Behav. Biomed. Mater. doi:10.1016/j.jmbbm.2014.06.008 Bhushan, B. & Koinkar, V.N. (1994) Nanoindentation hardness measurements using atomic force microscopy. Appl. Phys. Lett., 64, 1653–1655, American Institute of Physics. doi:10.1063/1.111949 Binnig, G., Quate, C.F. & Gerber, C. (1986) Atomic Force Microscope. Phys. Rev. Lett., 56, 930–933. doi:10.1103/PhysRevLett.56.930 Blank, S. & Dorf, B. (2013) The Startup Owner’s Manual: The Step-by-step Guide for Building a Great Company, John Wiley & Sons. Borodich, F.M. (2014) The Hertz-Type and Adhesive Contact Problems for Depth-Sensing Indentation. in Advances in Applied Mechanics. doi:10.1016/B978-0-12-800130-1.00003-5 Bos, E.J., Laan, K. Van Der, Helder, M.N., Mullender, M.G., Iannuzzi, D. & Zuijlen, P.P. Van. (2017) Noninvasive Measurement of Ear Cartilage Elasticity on the Cellular Level: A New Method to Provide Biomechanical Information for Tissue Engineering. Plast. Reconstr. Surg. - Glob. Open, 5, e1147, Wolters Kluwer Health. doi:10.1097/GOX.0000000000001147 Broitman, E. (2017, December 28) Indentation Hardness Measurements at Macro-, Micro-, and Nanoscale: A Critical Overview. Tribol. Lett., Springer New York LLC. doi:10.1007/s11249-016-0805-5 Brown, T.M. & Krishnamurthy, K. (2019) http://www.ncbi.nlm.nih.gov/pubmed/30570967

Histology,

Dermis.

StatPearls.

Retrieved

from

Bruin, D.M. de, Bremmer, R.H., Kodach, V.M., Kinkelder, R. de, Marle, J. van, Leeuwen, T.G. van & Faber, D.J. (2010) Optical phantoms of varying geometry based on thin building blocks with controlled optical properties. J. Biomed. Opt., 15, 025001. doi:10.1117/1.3369003 Brun, E. (2012) Ambiguity: A Useful Component of “Fuzziness” in Innovation. in IFIP Advances in Information and Communication Technology, pp. 412–424. doi:10.1007/978-3-642-33980-6_46 Brusselaers, N., Pirayesh, A., Hoeksema, H., Verbelen, J., Blot, S. & Monstrey, S. (2010) Burn scar assessment: A systematic review of objective scar assessment tools. Burns, 36, 1157–1164. doi:10.1016/j.burns.2010.03.016 Burnham, N.A., Chen, X., Hodges, C.S., Matei, G.A., Thoreson, E.J., Roberts, C.J., Davies, M.C., et al. (2003) Comparison of calibration methods for atomic-force microscopy cantilevers. Nanotechnology, 14, 1–6. doi:10.1088/0957-4484/14/1/301 Buzio, R., Boragno, C., Biscarini, F., Buatier De Mongeot, F. & Valbusa, U. (2003) The contact mechanics of fractal surfaces. Nat. Mater., 2, 233–236. doi:10.1038/nmat855 Campbell, B., Dobson, L., Higgins, J., Dillon, B., Marlow, M. & Pomfrett, C. (2017) A new health technology assessment system for devices: The first five years. Int. J. Technol. Assess. Health Care. doi:10.1017/S0266462317000253 Cappella, B. (2016) Mechanical Properties of Polymers Measured through AFM Force-Distance Curves. Springer Laboratory, Cham: Springer International Publishing. doi:10.1007/978-3-319-29459-9 CB Insights. (2019) Why Startups Fail: Top 20 Reasons l CB Insights. CB Insights. Retrieved March 4, 2020, from https://www.cbinsights.com/research/startup-failure-reasons-top/ CCMO. (2020) Investigational Medical Device Dossier (IMDD). Retrieved June 11, 2020, from https://english.ccmo.nl/investigators/standard-research-file/d.-product-information/d2.-investigationalmedical-device-dossier-imdd Charitidis, C.A. (2011) Nanoscale Deformation and Nanomechanical Properties of Polydimethylsiloxane (PDMS). Ind. Eng. Chem. Res., 50, 565–570. doi:10.1021/ie100099g Chavan, D., Gruca, G., Man, S. De, Slaman, M., Rector, J.H., Heeck, K. & Iannuzzi, D. (2010) Ferrule-top atomic force microscope. Rev. Sci. Instrum., 81, 123702. doi:10.1063/1.3516044 Chavan, D., Watering, T.C. Van De, Gruca, G., Rector, J.H., Heeck, K., Slaman, M. & Iannuzzi, D. (2012) Ferrule-top nanoindenter: An optomechanical fiber sensor for nanoindentation. Rev. Sci. Instrum., 83, 115110.

152

References doi:10.1063/1.4766959 Chavan, D. (2014) Applications of Fiber-top Technology for Material Property Characterization at Nanoscale. Chavan, Dhwajal, Mo, J., Groot, M. de, Meijering, A., Boer, J.F. de & Iannuzzi, D. (2013) Collecting optical coherence elastography depth profiles with a micromachined cantilever probe. Opt. Lett., 38, 1476–8. doi:10.1364/OL.38.001476 Chin, L., Kennedy, B.F., Kennedy, K.M., Wijesinghe, P., Pinniger, G.J., Terrill, J.R., McLaughlin, R.A., et al. (2014) Three-dimensional optical coherence micro-elastography of skeletal muscle tissue. Biomed. Opt. Express, 5, 3090–102. doi:10.1364/BOE.5.003090 Christensen, R.M. (1972) Restrictions Upon Viscoelastic Relaxation Functions and Complex Moduli. Trans Soc Rheol, 16, 603–614. doi:10.1122/1.549265 Christensen, R.M. & Freund, L.B. (1984) Theory of Viscoelasticity, An Introduction, 2nd Edition. J. Appl. Mech., Vol. 51, Academic Press. doi:10.1115/1.3167591 Clarke, D.M. (1988) Case hardened steel-vickers hardness test. Retrieved May 14, 2020, from https://commons.wikimedia.org/wiki/File:Case_hardened_steel-vickers_hardness_test.png Clifford, C.A. & Seah, M.P. (2009) Improved methods and uncertainty analysis in the calibration of the spring constant of an atomic force microscope cantilever using static experimental methods. Meas. Sci. Technol., 20, 125501. doi:10.1088/0957-0233/20/12/125501 Cohen, S.R. & Kalfon-Cohen, E. (2013) Dynamic nanoindentation by instrumented nanoindentation and force microscopy: a comparative review. Beilstein J. Nanotechnol., 4, 815–833. doi:10.3762/bjnano.4.93 Constantinides, G., Ravi Chandran, K.S., Ulm, F.J. & Vliet, K.J. Van. (2006) Grid indentation analysis of composite microstructure and mechanics: Principles and validation. Mater. Sci. Eng. A, 430, 189–202. doi:10.1016/j.msea.2006.05.125 Corr, D.T. & Hart, D.A. (2013) Biomechanics of Scar Tissue and Uninjured Skin. Adv. Wound Care, 2, 37–43. doi:10.1089/wound.2011.0321 Cowin, S.C. & Doty, S.B. (2007) Tissue mechanics. Tissue Mech. doi:10.1007/978-0-387-49985-7 Cox, N.H. (2006) Palpation of the skin - An important issue. J. R. Soc. Med. doi:10.1258/jrsm.99.12.598 Cox, N.H. (2007) A literally blinded trial of palpation in dermatologic diagnosis. J. Am. Acad. Dermatol., 56, 949–951. doi:10.1016/j.jaad.2007.01.044 Crichton, M.L., Chen, X., Huang, H. & Kendall, M.A.F. (2013) Elastic modulus and viscoelastic properties of full thickness skin characterised at micro scales. Biomaterials, 34, 2087–2097, Elsevier Ltd. doi:10.1016/j.biomaterials.2012.11.035 Cropley, T.G. (2001) Andrews’ Diseases of the skin: Clinical dermatology. 9th ed. J. Am. Acad. Dermatol., 44, 882–883. doi:10.1016/s0190-9622(01)70200-9 Daly, C.H. (1982) Biomechanical properties of dermis. J. Invest. Dermatol., 79, 17–20. doi:10.1038/jid.1982.4 Danin, A., Georgesco, G., Touze, A. Le, Penaud, A., Quignon, R. & Zakine, G. (2012) Assessment of burned hands reconstructed with Integra® by ultrasonography and elastometry. Burns, 38, 998–1004. doi:10.1016/j.burns.2012.02.017 Delalleau, A., Josse, G., Lagarde, J.M., Zahouani, H. & Bergheau, J.M. (2008) A nonlinear elastic behavior to identify the mechanical parameters of human skin in vivo. Ski. Res. Technol., 14, 152–164. doi:10.1111/j.16000846.2007.00269.x Delalleau, Alexandre, Josse, G., Lagarde, J.M., Zahouani, H. & Bergheau, J.M. (2006) Characterization of the mechanical properties of skin by inverse analysis combined with the indentation test. J. Biomech., 39, 1603– 1610. doi:10.1016/j.jbiomech.2005.05.001 DeLoitte. (n.d.) Out of the valley of death How can entrepreneurs, corporations, and investors reinvigorate earlystage medtech innovation? Retrieved March 2, 2020, from https://www.advamed.org/sites/default/files/resource/medtech_innovation_report_2017.pdf Demirgüneş, B.K. (2015) Relative importance of perceived value, satisfaction and perceived risk on willingness to pay more. Int. Rev. Manag. Mark., 5, 211–220.

153

References Dennehy, D., Kasraian, L., O’raghallaigh, P. & Conboy, K. (2016) Product Market Fit Frameworks for Lean Product Development, Cambridge University. Diridollou, S., Berson, M., Vabre, V., Black, D., Karlsson, B., Auriol, F., Gregoire, J.M., et al. (1998) An in vivo method for measuring the mechanical properties of the skin using ultrasound. Ultrasound Med. Biol., 24, 215– 224. doi:10.1016/S0301-5629(97)00237-8 Diridollou, S., Patat, F., Gens, F., Vaillant, L., Black, D., Lagarde, J.M., Gall, Y., et al. (2000) In vivo model of the mechanical properties of the human skin under suction. Ski. Res. Technol., 6, 214–221. doi:10.1034/j.16000846.2000.006004214.x Dobrev, H. (2000) In vivo study of skin mechanical properties in psoriasis vulgaris. Acta Derm. Venereol., 80, 263–266. doi:10.1080/000155500750012135 Dobrev, Hristo. (2014) Novel Ideas: The Increased Skin Viscoelasticity - A Possible New Fifth Sign for the Very Early Diagnosis of Systemic Sclerosis. Curr. Rheumatol. Rev., 9, 261–267, Bentham Science Publishers Ltd. doi:10.2174/157339710904140417125455 Draaijers, L.J., Botman, Y.A.M., Tempelman, F.R.H., Kreis, R.W., Middelkoop, E. & Zuijlen, P.P.. van. (2004) Skin elasticity meter or subjective evaluation in scars: a reliability assessment. Burns, 30, 109–114. doi:10.1016/j.burns.2003.09.003 Ebenstein, D. (2010) Nanoindentation of Soft Tissues and Other Biological Materials. Handb. Nanoindentation, 279–324. doi:10.1201/b12116-10 Ebenstein, D.M. & Pruitt, L. a. (2006) Nanoindentation of Nanoindentation has recently emerged as a powerful tool for measuring, 1, 26–33. Edwards, C. & Marks, R. (1995) Evaluation of biomechanical properties of human skin. Clin. Dermatol., 13, 375–380. doi:10.1016/0738-081X(95)00078-T Egan, P., Sinko, R., LeDuc, P.R. & Keten, S. (2015) The role of mechanics in biological and bio-inspired systems. Nat. Commun., 6, 7418. doi:10.1038/ncomms8418 Elias, P.M. (2005, August) Stratum corneum defensive functions: An integrated view. J. Invest. Dermatol., Blackwell Publishing Inc. doi:10.1111/j.0022-202X.2005.23668.x Eliceiri, K., Conklin, M., Bredfeldt, J., Mackie, T., Liu, Y. & Keely, P. (2014) Automated quantification of aligned collagen for human breast carcinoma prognosis. J. Pathol. Inform. doi:10.4103/2153-3539.139707 Elsner, P., Berardesca, E., Wilhelm, K.-P., Berardesca, E. & Wilhelm, K.-P. (2001) Bioengineering of the Skin. Bioeng. Ski., CRC Press. doi:10.1201/b14261 Enomoto, D.N.H., Mekkes, J.R., Bossuyt, P.M.M., Hoekzema, R. & Bos, J.D. (1996) Quantification of cutaneous sclerosis with a skin elasticity meter in patients with generalized scleroderma. J. Am. Acad. Dermatol., 35, 381– 387, Mosby Inc. doi:10.1016/S0190-9622(96)90601-5 Es’haghian, S., Kennedy, K.M., Gong, P., Li, Q., Chin, L., Wijesinghe, P., Sampson, D.D., et al. (2017) In vivo volumetric quantitative micro-elastography of human skin. Biomed. Opt. Express, 8, 2458. doi:10.1364/BOE.8.002458 Escoffier, C., Rigal, J. de, Rochefort, A., Vasselet, R., Lévêque, J.L. & Agache, P.G. (1989) Age-related mechanical properties of human skin: an in vivo study. J. Invest. Dermatol. EU Parliament. (2017) Regulation (EU) 2017/745 of the European Parliament of the Council of 5 April 2017 on medical devices amending Directive 2001/83/EC. Retrieved from http://eur-lex.europa.eu/legalcontent/EN/TXT/?uri=OJ:L:2017:117:TOC European Commission. (2015) G . Technology readiness levels (TRL). Horizon 2020. Retrieved March 3, 2020, from https://ec.europa.eu/research/participants/data/ref/h2020/wp/2014_2015/annexes/h2020-wp1415-annexg-trl_en.pdf F.

Fernando, G. (2008) Introduction doi:10.1016/j.matchar.2007.05.008

polymer

viscoelasticity.

Mater.

Charact.,

Vol.

59.

Farine, M. (2013) Instrumented Indentation of Soft Materials and Biological Tissues. Fercher, A.F., Drexler, W., Hitzenberger, C.K. & Lasser, T. (2003) Optical coherence tomography - principles and applications. Reports Prog. Phys., 66, 239, IOP Publishing. doi:10.1088/0034-4885/66/2/204

154

References Firooz, A., Sadr, B., Babakoohi, S., Sarraf-Yazdy, M., Fanian, F., Kazerouni-Timsar, A., Nassiri-Kashani, M., et al. (2012) Variation of Biophysical Parameters of the Skin with Age, Gender, and Body Region. Sci. World J., 2012, 1–5. doi:10.1100/2012/386936 Fong, S.S.L., Hung, L.K. & Cheng, J.C.Y. (1997) The cutometer and ultrasonography in the assessment of postburn hypertrophic scar: A preliminary study. Burns, Vol. 23, pp. S12–S18. doi:10.1016/S03054179(96)00095-2 Fung, Y.C. & Skalak, R. (1982) Biomechanics. Mechanical Properties of Living Tissues. J. Appl. Mech. doi:10.1115/1.3162171 Gambichler, T., Jaedicke, V. & Terras, S. (2011, September 7) Optical coherence tomography in dermatology: Technical and clinical aspects. Arch. Dermatol. Res., Springer. doi:10.1007/s00403-011-1152-x Gavara, N. (2016) Combined strategies for optimal detection of the contact point in AFM force-indentation curves obtained on thin samples and adherent cells. Sci. Rep., 6, 21267, Nature Publishing Group. doi:10.1038/srep21267 Geerligs, M. (2010) , Skin layer mechanics. Ski. layer Mech., Eindhoven, NL, Technische Universiteit Eindhoven. Graham, H.K., McConnell, J.C., Limbert, G. & Sherratt, M.J. (2019) How stiff is skin? Exp. Dermatol. doi:10.1111/exd.13826 Grant, R.M. (2000) Contemporary doi:10.1108/ws.2000.07949aae.002

Strategy

Analysis

(3rd

ed.).

Work

Study,

Vol.

49.

Grover, S. & Grewal, R. (2008) Fitzpatrick’s Dermatology in General Medicine. Med. J. Armed Forces India. doi:10.1016/s0377-1237(08)80108-6 Guo, C.L., Harris, N.C., Wijeratne, S.S., Frey, E.W. & Kiang, C.H. (2013) Multiscale mechanobiology: Mechanics at the molecular, cellular, and tissue levels. Cell Biosci., 3, 25. doi:10.1186/2045-3701-3-25 Gutierrez-Lemini, D. (2014) Engineering viscoelasticity. Eng. Viscoelasticity, 2139, 1–353. doi:10.1007/978-14614-8139-3 Guy, R.H. (1996) Handbook of non-invasive methods and the skin. J. Control. Release, 42, 102. doi:10.1016/0168-3659(96)01365-X Han, L., Frank, E.H., Greene, J.J., Lee, H.-Y., Hung, H.-H.K., Grodzinsky, A.J. & Ortiz, C. (2011) TimeDependent Nanomechanics of Cartilage. Biophys. J., 100, 1846–1854. doi:10.1016/j.bpj.2011.02.031 Heijden, J.P. van der. (2013) PhD Thesis |Telemedicine in dermatology: Evaluation of secondary and tertiary teledermatology. Retrieved from https://dare.uva.nl/search?identifier=9f6c05a8-fb07-4832-a3c2be81b6035fe5 Hendriks, F.M., Brokken, D., Eemeren, J.T.W.M. van, Oomens, C.W.J., Baaijens, F.P.T. & Horsten, J.B.A.M. (2003) A numerical-experimental method to characterize the non-linear mechanical behavior of human skin. Ski. Res. Technol., 9, 274–283. doi:10.1034/j.1600-0846.2003.00019.x Hendriks, F.M., Brokken, D., Oomens, C.W.J. & Baaijens, F.P.T. (2004) Influence of hydration and experimental length scale on the mechanical response of human skin in vivo, using optical coherence tomography. Ski. Res. Technol., 10, 231–241. doi:10.1111/j.1600-0846.2004.00077.x Hendriks, F.M., Brokken, D., Oomens, C.W.J., Bader, D.L. & Baaijens, F.P.T. (2006) The relative contributions of different skin layers to the mechanical behavior of human skin in vivo using suction experiments. Med. Eng. Phys. doi:10.1016/j.medengphy.2005.07.001 Herbert, E.G., Oliver, W.C. & Pharr, G.M. (2008) Nanoindentation and the dynamic characterization of viscoelastic solids. J. Phys. D. Appl. Phys., 41, 074021. doi:10.1088/0022-3727/41/7/074021 Hertz, H. (1882) ‘Uber die beruhrung fester elastische korper und uber die harte (On the contact of rigid elastic solids and on hardness).’ Verhandlungen des Vereins zur Beforderung des Gewerbefleisses, Leipzig. Heymans, N. (2003) Constitutive equations for polymer viscoelasticity derived from hierarchical models in cases of failure of time-temperature superposition. Signal Processing, 83, 2345–2357. doi:10.1016/S01651684(03)00187-7 Hinterhuber, A. (2008) Customer value-based pricing strategies: Why companies resist. J. Bus. Strategy, 29, 41– 50, Emerald Group Publishing Limited. doi:10.1108/02756660810887079

155

References Hisyam A. Razak, A., Szabo, P. & Skov, A.L. (2015) Enhancement of dielectric permittivity by incorporating PDMS-PEG multiblock copolymers in silicone elastomers. RSC Adv., 5, 53054–53062. doi:10.1039/C5RA09708H Holbrook, K.A. & Odland, G.F. (1974) Regional Differences in the Thickness (Cell Layers) of the Human Stratum Corneum: An Ultrastructural Analysis. J. Invest. Dermatol., 62, 415–422. doi:10.1111/1523-1747.ep12701670 Holzapfel, G.A. & Ogden, R.W. (2006) Mechanics of biological tissue. Mech. Biol. Tissue. doi:10.1007/3-54031184-X Honari, G. & Maibach, H. (2014) Skin Structure and Function. in Applied Dermatotoxicology, pp. 1–10, Elsevier. doi:10.1016/B978-0-12-420130-9.00001-3 Hoorn, H. van, Kurniawan, N.A., Koenderink, G.H. & Iannuzzi, D. (2016) Local dynamic mechanical analysis for heterogeneous soft matter using ferrule-top indentation. Soft Matter, 12, 3066–3073, Royal Society of Chemistry. doi:10.1039/C6SM00300A Humphrey, J.D. (2003) Continuum biomechanics of soft biological tissues. Proc. R. Soc. A Math. Phys. Eng. Sci. doi:10.1098/rspa.2002.1060 Hussain, S.H., Limthongkul, B. & Humphreys, T.R. (2013) The Biomechanical Properties of the Skin. Dermatologic Surg., 39, 193–203. doi:10.1111/dsu.12095 Iannuzzi, D. (2017) Entrepreneurship for Physicists A practical guide to move inventions from university to market. Entrep. Phys. A Pract. Guid. to move Invent. from Univ. to Mark. doi:10.1088/978-1-6817-4668-5 Iannuzzi, D., Gruca, G. & Sven, D.M. (2011) Optical fiber, method of preparation thereof and device. Retrieved from https://patents.google.com/patent/WO2011031142A1/en?inventor=Davide+Iannuzzi Jakes, J.E. & Stone, D.S. (2011) The edge effect in nanoindentation. Philos. Mag., 91, 7–9. doi:10.1080/14786435.2010.495360 Jaspers, M.E.H., Feroldi, F., Vlig, M., Boer, J.F. de & Zuijlen, P.P.M. van. (2017) In vivo polarization-sensitive optical coherence tomography of human burn scars: birefringence quantification and correspondence with histologically determined collagen density. J. Biomed. Opt., 22, 1. doi:10.1117/1.jbo.22.12.121712 Jelen, C., Mattei, G., Montemurro, F., Maria, C. De, Mattioli-Belmonte, M. & Vozzi, G. (2013) Bone scaffolds with homogeneous and discrete gradient mechanical properties. Mater. Sci. Eng. C, 33, 28–36, Elsevier B.V. doi:10.1016/j.msec.2012.07.046 Jelvehgaran, P., Alderliesten, T., Weda, J.J.A., Bruin, M. De, Faber, D.J., Hulshof, M.C.C.M., Leeuwen, T.G. Van, et al. (2017) Visibility of fiducial markers used for image-guided radiation therapy on optical coherence tomography for registration with CT: An esophageal phantom study: An. Med. Phys. doi:10.1002/mp.12624 Jiang, D., Feng, X., Qu, B., Wang, Y. & Fang, D. (2012) Rate-dependent interaction between thin films and interfaces during micro/nanoscale transfer printing. Soft Matter, 8, 418–423. doi:10.1039/C1SM06483E Jiang, W.G., Su, J.J. & Feng, X.Q. (2008) Effect of surface roughness on nanoindentation test of thin films. Eng. Fract. Mech., 75, 4965–4972. doi:10.1016/j.engfracmech.2008.06.016 Johnson, K.L. (1987) Contact Mechanics. Handouts, 452. doi:10.1115/1.3261297 Johnson, K.L., Kendall, K. & Roberts, A.D. (1971) Surface energy and the contact of elastic solids. Proc. R. Soc. London. A. Math. Phys. Sci., 324, 301–313. doi:10.1098/rspa.1971.0141 Joodaki, H. & Panzer, M.B. (2018) Skin mechanical properties and modeling: A review. Proc. Inst. Mech. Eng. Part H J. Eng. Med. doi:10.1177/0954411918759801 Jor, J.W.Y., Parker, M.D., Taberner, A.J., Nash, M.P. & Nielsen, P.M.F. (2013) Computational and experimental characterization of skin mechanics: Identifying current challenges and future directions. Wiley Interdiscip. Rev. Syst. Biol. Med., 5, 539–556. doi:10.1002/wsbm.1228 Kanitakis, J. (2002) Anatomy, histology and immunohistochemistry of normal human skin. Eur. J. Dermatology. Katz, J.L., Misra, A., Spencer, P., Wang, Y., Bumrerraj, S., Nomura, T., Eppell, S.J., et al. (2007) Multiscale mechanics of hierarchical structure/property relationships in calcified tissues and tissue/material interfaces. Mater. Sci. Eng. C, 27, 450–468. doi:10.1016/j.msec.2006.05.055 Kauer, M., Vuskovic, V., Dual, J., Szekely, G. & Bajka, M. (2001) Inverse finite element characterization of soft

156

References tissues. Lect. Notes Comput. Sci. (including Subser. Lect. Notes Artif. Intell. Lect. Notes Bioinformatics). doi:10.1007/3-540-45468-3_16 Kaufman, J.D. & Klapperich, C.M. (2009) Surface detection errors cause overestimation of the modulus in nanoindentation on soft materials. J. Mech. Behav. Biomed. Mater., 2, 312–317. doi:10.1016/j.jmbbm.2008.08.004 Kaufman, J.D., Miller, G.J., Morgan, E.F. & Klapperich, C.M. (2008) Time-dependent mechanical characterization of poly(2-hydroxyethyl methacrylate) hydrogels using nanoindentation and unconfined compression. J. Mater. Res., 23, 1472–1481. doi:10.1557/JMR.2008.0185 Kennedy, Brendan F., McLaughlin, R.A., Kennedy, K.M., Chin, L., Curatolo, A., Tien, A., Latham, B., et al. (2014) Optical coherence micro-elastography: mechanical-contrast imaging of tissue microstructure. Biomed. Opt. Express, 5, 2113. doi:10.1364/boe.5.002113 Kennedy, Brendan F., Wijesinghe, P. & Sampson, D.D. (2017) The emergence of optical elastography in biomedicine. Nat. Photonics, 11, 215–221. doi:10.1038/nphoton.2017.6 Kennedy, Brendan Francis, Kennedy, K.M. & Sampson, D.D. (2014) A review of optical coherence elastography: Fundamentals, techniques and prospects. IEEE J. Sel. Top. Quantum Electron., Vol. 20, pp. 272–288. doi:10.1109/JSTQE.2013.2291445 Kennedy, K.M., Chin, L., McLaughlin, R.A., Latham, B., Saunders, C.M., Sampson, D.D. & Kennedy, B.F. (2015) Quantitative micro-elastography: imaging of tissue elasticity using compression optical coherence elastography. Sci. Rep., 5, 15538. doi:10.1038/srep15538 Kennedy, K.M., Kennedy, B.F., McLaughlin, R.A. & Sampson, D.D. (2012) Needle optical coherence elastography for tissue boundary detection. Opt. Lett., 37, 2310. doi:10.1364/OL.37.002310 Kennedy, K.M., McLaughlin, R. a, Kennedy, B.F., Tien, A., Latham, B., Saunders, C.M. & Sampson, D.D. (2013) Needle optical coherence elastography for the measurement of microscale mechanical contrast deep within human breast tissues. J. Biomed. Opt., 18, 121510. doi:10.1117/1.JBO.18.12.121510 Khalil, A.S., Chan, R.C., Chau, A.H., Bouma, B.E. & Mofrad, M.R.K. (2005) Tissue elasticity estimation with optical coherence elastography: Toward mechanical characterization of in vivo soft tissue. Ann. Biomed. Eng., 33, 1631–1639. doi:10.1007/s10439-005-6766-3 Khanafer, K., Duprey, A., Schlicht, M. & Berguer, R. (2009) Effects of strain rate, mixing ratio, and stress-strain definition on the mechanical behavior of the polydimethylsiloxane (PDMS) material as related to its biological applications. Biomed. Microdevices, 11, 503–508. doi:10.1007/s10544-008-9256-6 Khatyr, F., Imberdis, C., Varchon, D., Lagarde, J.M. & Josse, G. (2006) Measurement of the mechanical properties of the skin using the suction test. Ski. Res. Technol., 12, 24–31. doi:10.1111/j.0909725X.2006.00126.x Kim, Y.-C., Gwak, E.-J., Ahn, S., Kang, N.-R., Han, H.N., Jang, J. & Kim, J.-Y. (2018) Indentation size effect for spherical nanoindentation on nanoporous gold. Scr. Mater., 143, 10–14. doi:10.1016/j.scriptamat.2017.09.002 Kirk, R.W., Kennedy, B.F., Sampson, D.D. & McLaughlin, R.A. (2015) Near Video-Rate Optical Coherence Elastography by Acceleration With a Graphics Processing Unit. J. Light. Technol., 33, 3481–3485. doi:10.1109/JLT.2015.2413402 Kirkpatrick, S.J., Wang, R.K., Duncan, D.D., Kulesz-Martin, M. & Lee, K. (2006) Imaging the mechanical stiffness of skin lesions by in vivo acousto-optical elastography. Opt. Express, 14, 9770. doi:10.1364/oe.14.009770 Kontomaris, S. V. & Malamou, A. (2020) Hertz model or Oliver & Pharr analysis? Tutorial regarding AFM nanoindentation experiments on biological samples. Mater. Res. Express, 7, 033001. doi:10.1088/20531591/ab79ce Lakes, R. (2009) Viscoelastic materials. Viscoelastic Mater., Vol. 9780521885, Cambridge University Press. doi:10.1017/CBO9780511626722 Larin, K. V. & Sampson, D.D. (2017) Optical coherence elastography – OCT at work in tissue biomechanics [Invited]. Biomed. Opt. Express, 8, 1172. doi:10.1364/boe.8.001172 Lavet, C. & Ammann, P. (2017) Osteoarthritis like Alteration of Cartilage and Subchondral Bone Induced by

157

References Protein Malnutrition is treated by Nutritional Essential Amino Acids Supplements. Osteoarthr. Cartil., 25, S293. doi:10.1016/j.joca.2017.02.495 Lei, F. & Szeri, A.Z. (2007) Inverse analysis of constitutive models: Biological soft tissues. J. Biomech., 40, 936– 940, Elsevier. doi:10.1016/j.jbiomech.2006.03.014 Lekka, M., Gil, D., Pogoda, K., Dulińska-Litewka, J., Jach, R., Gostek, J., Klymenko, O., et al. (2012) Cancer cell detection in tissue sections using AFM. Arch. Biochem. Biophys., 518, 151–156. doi:10.1016/j.abb.2011.12.013 Leonardo, P. (2010) Inverse determination of viscoelastic properties of human fingertip skin. RMZ – Mater. Geoenvironment. Leta, F.R., Gomes, J.F.S., Costa, P.B. & O. Baldner, F. De. (2015) Mechanical and Materials Engineering of Modern Structure and Component Design. (A. Öchsner & H. Altenbach, Eds.)Adv. Struct. Mater. Advanced Structured Materials, Vol. 70, Cham: Springer International Publishing. doi:10.1007/978-3-319-19443-1 Levecq, O., Davis, A.J., Perrot, J.L., Dubois, A., Siret, D. & Azimani, H. (2018) Advances in optical coherence tomography for dermatology. Unconv. Opt. Imaging eds. Fournier, C., Georges, M.P. & Popescu, G., p. 19, SPIE. doi:10.1117/12.2315765 Li, J., Feroldi, F., Lange, J. de, Daniels, J.M.A., Grünberg, K. & Boer, J.F. de. (2015) Polarization sensitive optical frequency domain imaging system for endobronchial imaging. Opt. Express, 23, 3390, Optical Society of America. doi:10.1364/OE.23.003390 Li, J., Groot, M. de, Helderman, F., Mo, J., Daniels, J.M.A., Grünberg, K., Sutedja, T.G., et al. (2012) High speed miniature motorized endoscopic probe for optical frequency domain imaging. Opt. Express, 20, 24132, Optical Society of America. doi:10.1364/oe.20.024132 Li, X. & Bhushan, B. (2002) A review of nanoindentation continuous stiffness measurement technique and its applications. Mater. Charact., 48, 11–36, Elsevier Inc. doi:10.1016/S1044-5803(02)00192-4 Liang, X. (2013) Biomechanical Properties of In Vivo Human Skin From Dynamic Optical Coherence Elastography. IEEE Trans Biomed Eng., 57, 953–959. doi:10.1109/TBME.2009.2033464.Biomechanical Lim, C.T., Zhou, E.H. & Quek, S.T. (2006, January) Mechanical models for living cells - A review. J. Biomech. doi:10.1016/j.jbiomech.2004.12.008 Lin, D.C. & Horkay, F. (2008) Nanomechanics of polymer gels and biological tissues: A critical review of analytical approaches in the Hertzian regime and beyond. Soft Matter. doi:10.1039/b714637j Lokody, I. (2014) Microenvironment: Tumour-promoting tissue mechanics. Nat. Rev. Cancer, 14, 296–297. doi:10.1038/nrc3727 Lopez-Higuera, Jose Miguel; Culshaw, B. (2008) Engineering a high-tech business: Entrepreneurial experiences and insights, SPIE Press. Mammoto, T., Mammoto, A. & Ingber, D.E. (2013) Mechanobiology and Developmental Control. Annu. Rev. Cell Dev. Biol., 29, 27–61. doi:10.1146/annurev-cellbio-101512-122340 Mark, J.E. (2009) Polymer Data Handbook, 2nd ed. Polymer Data Handbook, 2nd ed . Edited by James E. Mark (University of Cincinnati, OH). Oxford University Press: New York. 2009 . xii + 1250 pp. $195.00. ISBN 9780-19-518101-2 . J. Am. Chem. Soc., 131, 16330–16330. doi:10.1021/ja907879q Maslow, A.H. (1966) The psychology of science; a reconnaissance. John Dewey Soc. Lecturesh. Ser. doi:EB NC MASL Mattei, G. & Ahluwalia, A. (2019) A new analytical method for estimating lumped parameter constants of linear viscoelastic models from strain rate tests. Mech. Time-Dependent Mater., 23, 327–335. doi:10.1007/s11043018-9385-0 Mattei, G., Gruca, G., Rijnveld, N. & Ahluwalia, A. (2015) The nano-epsilon dot method for strain rate viscoelastic characterisation of soft biomaterials by spherical nano-indentation. J. Mech. Behav. Biomed. Mater., 50, 150–159. doi:10.1016/j.jmbbm.2015.06.015 Mattei, G., Tirella, A., Gallone, G. & Ahluwalia, A. (2014) Viscoelastic characterisation of pig liver in unconfined compression. J. Biomech., 47, 2641–2646. doi:10.1016/j.jbiomech.2014.05.017 Mattei, Giorgio & Ahluwalia, A. (2016) Sample, testing and analysis variables affecting liver mechanical

158

References properties: A review. Acta Biomater. doi:10.1016/j.actbio.2016.08.055 Mattei, Giorgio, Cacopardo, L. & Ahluwalia, A. (2017) Micro-mechanical viscoelastic properties of crosslinked hydrogels using the nano-epsilon dot method. Materials (Basel)., 10, 889. doi:10.3390/ma10080889 Mattei, Giorgio, Ferretti, C., Tirella, A., Ahluwalia, A. & Mattioli-Belmonte, M. (2015) Decoupling the role of stiffness from other hydroxyapatite signalling cues in periosteal derived stem cell differentiation. Sci. Rep., 5, 10778. doi:10.1038/srep10778 Mattei, Giorgio, Magliaro, C., Giusti, S., Ramachandran, S.D., Heinz, S., Braspenning, J. & Ahluwalia, A. (2017) On the adhesion-cohesion balance and oxygen consumption characteristics of liver organoids. (N. Cordes, Ed.)PLoS One, 12, e0173206. doi:10.1371/journal.pone.0173206 Maugis, D. (1992) Adhesion of spheres: The JKR-DMT transition using a dugdale model. J. Colloid Interface Sci., 150, 243–269. doi:10.1016/0021-9797(92)90285-T Mencik, J., He, L.H. & Swain, M. V. (2009) Determination of viscoelastic-plastic material parameters of biomaterials by instrumented indentation. J. Mech. Behav. Biomed. Mater., 2, 318–325. doi:10.1016/j.jmbbm.2008.09.002 Meskell, M. (2010) Principles of Anatomy and Physiology. J. Anat., Vol. 217, Wiley. doi:10.1111/j.14697580.2010.01292.x Miguel, E., D’Angelo, M.L., Cannella, F., Bianchi, M., Memeo, M., Bicchi, A., Caldwell, D.G., et al. (2015) Characterization of nonlinear finger pad mechanics for tactile rendering. 2015 IEEE World Haptics Conf., pp. 63–68, IEEE. doi:10.1109/WHC.2015.7177692 Moshtagh, P.R., Pouran, B., Korthagen, N.M., Zadpoor, A.A. & Weinans, H. (2016) Guidelines for an optimized indentation protocol for measurement of cartilage stiffness: The effects of spatial variation and indentation parameters. J. Biomech., 49, 3602–3607. doi:10.1016/j.jbiomech.2016.09.020 Müller, W.E.G., Tolba, E., Schröder, H.C., Wang, S., Glaßer, G., Muñoz-Espí, R., Link, T., et al. (2015) A new polyphosphate calcium material with morphogenetic activity. Mater. Lett., 148, 163–166. doi:10.1016/j.matlet.2015.02.070 Mulligan, J.A., Untracht, G.R., Chandrasekaran, S.N., Brown, C.N. & Adie, S.G. (2016) Emerging Approaches for High-Resolution Imaging of Tissue Biomechanics With Optical Coherence Elastography. IEEE J. Sel. Top. Quantum Electron., 22, 1–20. doi:10.1109/JSTQE.2015.2481705 Neufurth, M., Wang, X., Schröder, H.C., Feng, Q., Diehl-Seifert, B., Ziebart, T., Steffen, R., et al. (2014) Engineering a morphogenetically active hydrogel for bioprinting of bioartificial tissue derived from human osteoblast-like SaOS-2 cells. Biomaterials, 35, 8810–8819. doi:10.1016/j.biomaterials.2014.07.002 Ní Annaidh, A., Bruyère, K., Destrade, M., Gilchrist, M.D. & Otténio, M. (2012) Characterization of the anisotropic mechanical properties of excised human skin. J. Mech. Behav. Biomed. Mater., 5, 139–148. doi:10.1016/j.jmbbm.2011.08.016 Nicoletti, G., Brenta, F., Bleve, M., Pellegatta, T., Malovini, A., Faga, A. & Perugini, P. (2015) Long-term in vivo assessment of bioengineered skin substitutes: A clinical study. J. Tissue Eng. Regen. Med., 9, 460–468. doi:10.1002/term.1939 Niu, T., Cao, G. & Xiong, C. (2016) Fracture behavior of graphene mounted on stretchable substrate. Carbon N. Y., 109, 852–859. doi:10.1016/j.carbon.2016.08.087 Oliver, W.C. & Pharr, G.M. (1992) An improved technique for determining hardness and elastic modulus using load and displacement sensing indentation experiments. J. Mater. Res., 7, 1564–1583. doi:10.1557/JMR.1992.1564 Oliver, W.C. & Pharr, G.M. (2004) Measurement of hardness and elastic modulus by instrumented indentation: Advances in understanding and refinements to methodology. J. Mater. Res., 19, 3–20. doi:10.1557/jmr.2004.19.1.3 Oxlund, H., Manschot, J. & Viidik, A. (1988) The role of elastin in the mechanical properties of skin. J. Biomech., 21, 213–218. doi:10.1016/0021-9290(88)90172-8 Oyen, M. L. (2006) Analytical techniques for indentation of viscoelastic materials. Philos. Mag., 86, 5625–5641. doi:10.1080/14786430600740666 Oyen, M. L. (2013) Nanoindentation of biological and biomimetic materials. Exp. Tech. doi:10.1111/j.1747-

159

References 1567.2011.00716.x Oyen, Michelle L. (2005) Spherical Indentation Creep Following Ramp Loading. J. Mater. Res., 20, 2094–2100. doi:10.1557/JMR.2005.0259 Páez-Avilés, Cristina, Esteve Juanola-Feliu, Islam Bogachan-Tahirbegi, Mónica Mir, Manel González-Piñero, and J.S. (2015) Innovation And Technology Transfer Of Medical Devices Fostered By Cross-Disciplinary Communities Of Practitioners. Int. J. Innov. Manag., 19, 1540012. doi:10.1142/S1363919615400125 Pakshir, P. & Hinz, B. (2018) The big five in fibrosis: Macrophages, myofibroblasts, matrix, mechanics, and miscommunication. Matrix Biol. doi:10.1016/j.matbio.2018.01.019 Pedersen, L., Hansen, B. & Jemec, G.B.E. (2003) Mechanical properties of the skin: A comparison between two suction cup methods. Ski. Res. Technol., 9, 111–115. doi:10.1034/j.1600-0846.2003.00021.x Pedersen, L. & Jemec, G. (2006) Mechanical Properties and Barrier Function of Healthy Human Skin. Acta Derm. Venereol., 86, 308–311. doi:10.2340/00015555-0080 Pensalfini, M., Weickenmeier, J., Rominger, M., Santoprete, R., Distler, O. & Mazza, E. (2018) Location-specific mechanical response and morphology of facial soft tissues. J. Mech. Behav. Biomed. Mater., 78, 108–115. doi:10.1016/j.jmbbm.2017.10.021 Pierard, G.E., Nikkels-Tassoudji, N. & Pierard-Franchimont, C. (1995) Influence of the test area on the mechanical properties of skin. Dermatology. doi:10.1159/000246472 Pittenger, B. & Scientist, S.A. (2013) Advances in AFM Nanomechanics AFM provides High resolution mapping of various sample properties. Retrieved from https://www.bruker.com/fileadmin/user_upload/8-PDFDocs/SurfaceAnalysis/AFM/Webinars/Advances_in_AFM_Nanomechanics-Feb20130AFMwebinarslides.pdf Qiu, Y., Wang, Y., Xu, Y., Chandra, N., Haorah, J., Hubbi, B., Pfister, B.J., et al. (2016) Quantitative optical coherence elastography based on fiber-optic probe for in situ measurement of tissue mechanical properties. Biomed. Opt. Express, 7, 688. doi:10.1364/boe.7.000688 Rodrigues, L. (2001) EEMCO Guidance to the in vivo Assessment of Tensile Functional Properties of the Skin. Skin Pharmacol. Physiol., 14, 52–67. doi:10.1159/000056334 Rogers, E.M., Singhal, A. & Quinlan, M.M. (2019) Diffusion of Innovations 1. in An Integrated Approach to Communication Theory and Research, pp. 415–434, Routledge. doi:10.4324/9780203710753-35 Ryssel, H., Germann, G., Kloeters, O., Gazyakan, E. & Radu, C.A. (2010) Dermal substitution with Matriderm® in burns on the dorsum of the hand. Burns. doi:10.1016/j.burns.2010.05.003 SALTER, D.C., McARTHUR, H.C., CROSSE, J.E. & DICKENS, A.D. (1993) Skin mechanics measured in vivo using torsion: a new and accurate model more sensitive to age, sex and moisturizing treatment. Int. J. Cosmet. Sci., 15, 200–218. doi:10.1111/j.1467-2494.1993.tb00075.x Sarasvathy, S.D. (2001) What Makes Entrepreneurs Entrepreneurial? Bus. Soc. Rev., 18. Retrieved from http://search.ebscohost.com/login.aspx?direct=true&db=buh&AN=5658599&site=ehost-live Sarasvathy, S.D. (2008) Effectuation: Elements of entrepreneurial expertise. Eff. Elem. Entrep. Expert. doi:10.4337/9781848440197 Sarker, B., Singh, R., Zehnder, T., Forgber, T., Alexiou, C., Cicha, I., Detsch, R., et al. (2017) Bioactive and Compatible Polymers Macromolecular interactions in alginate–gelatin hydrogels regulate the behavior of human fibroblasts. J. Bioact. Compat. Polym., 32, 309–324. doi:10.1177/0883911516668667 Schmitt, J. (1998) OCT elastography: imaging microscopic deformation and strain of tissue. Opt. Express, 3, 199–211. doi:10.1364/OE.3.000199 Shaheen, A.A.M., Wan, A.F. & Myers, R.P. (2007) FibroTest and FibroScan for the Prediction of Hepatitis CRelated Fibrosis: A Systematic Review of Diagnostic Test Accuracy. Am. J. Gastroenterol., 102, 2589–2600. doi:10.1111/j.1572-0241.2007.01466.x Shaw, M.C. & deSalvo, G.J. (1970) A new approach to plasticity and its application to blunt two dimensional indenters. J. Manuf. Sci. Eng. Trans. ASME, 92, 469–479. doi:10.1115/1.3427785 Shaw, M.T. & MacKnight, W.J. (2005) Introduction to Polymer Viscoelasticity: Third Edition. Introd. to Polym. Viscoelasticity Third Ed. doi:10.1002/0471741833

160

References Shen, L., Liu, T. & Lv, P. (2005) Polishing effect on nanoindentation behavior of nylon 66 and its nanocomposites. Polym. Test., 24, 746–749. doi:10.1016/j.polymertesting.2005.04.004 Strojny, A. & Gerberich, W.W. (1998) Experimental analysis of viscoelastic behavior in nanoindentation. Mater. Res. Soc. Symp. - Proc., 522, 159–164. doi:10.1557/proc-522-159 Suarez, F. & Lanzolla, G. (2005) The half-truth of first-mover advantage. Harv. Bus. Rev., 83, 121–127. Swanson, E.A. & Fujimoto, J.G. (2017) The ecosystem that powered the translation of OCT from fundamental research to clinical and commercial impact [Invited]. Biomed. Opt. Express, 8, 1638, The Optical Society. doi:10.1364/boe.8.001638 Tearney, G.J., Brezinski, M.E., Bouma, B.E., Hee, M.R., Southern, J.F. & Fujimoto, J.G. (1995) Determination of the refractive index of highly scattering human tissue by optical coherence tomography. Opt. Lett. doi:10.1364/ol.20.002258 Tirella, A., Mattei, G. & Ahluwalia, A. (2014) Strain rate viscoelastic analysis of soft and highly hydrated biomaterials. J. Biomed. Mater. Res. - Part A, 102, 3352–3360. doi:10.1002/jbm.a.34914 Tirella, Annalisa, Marca, M. La, Brace, L.A., Mattei, G., Aylott, J.W. & Ahluwalia, A. (2015) Nano-in-Micro Self-Reporting Hydrogel Constructs. J. Biomed. Nanotechnol., 11, 1451–1460. doi:10.1166/jbn.2015.2085 Transparency Market Research. (n.d.) Dermatology Diagnostic Devices and Therapeutics Market Size, Share & Trend | Industry Analysis Report, 2024. Retrieved March 2, 2020, from https://www.transparencymarketresearch.com/dermatology-diagnostic-devices-therapeutics-market.html Tweedie, C.A. & Vliet, K.J. Van. (2006) Contact creep compliance of viscoelastic materials via nanoindentation. J. Mater. Res., 21, 1576–1589. doi:10.1557/jmr.2006.0197 Ulrich, M., Themstrup, L., Carvalho, N. De, Manfredi, M., Grana, C., Ciardo, S., Kästle, R., et al. (2016) Dynamic Optical Coherence Tomography in Dermatology. Dermatology, 232, 298–311, S. Karger AG. doi:10.1159/000444706 VanLandingham, M.R. (2003) Review of instrumented indentation. J. Res. Natl. Inst. Stand. Technol., 108, 249– 265, National Institute of Standards and Technology. doi:10.6028/jres.108.024 Vashaghian, M., Ruiz-Zapata, A.M., Kerkhof, M.H., Zandieh-Doulabi, B., Werner, A., Roovers, J.P. & Smit, T.H. (2017) Toward a new generation of pelvic floor implants with electrospun nanofibrous matrices: A feasibility study. Neurourol. Urodyn., 36, 565–573. doi:10.1002/nau.22969 Verhaegen, P.D.H.M., Wal, M.B.A. Van Der, Middelkoop, E. & Zuijlen, P.P.M. Van. (2011) Objective scar assessment tools: A clinimetric appraisal. Plast. Reconstr. Surg., 127, 1561–1570. doi:10.1097/PRS.0b013e31820a641a Vogel, H.G. (1970) Tensile strength of skin wounds in rats after treatment with corticosteroids. Acta Endocrinol. (Copenh)., 64, 295–303. doi:10.1530/acta.0.0640295 Vogel, H.G. (1987) Age dependence of mechanical and biochemical properties of human skin. Part II: Hysteresis, relaxation, creep and repeated strain experiments. Bioeng. Ski., 3, 141–176. Vogt, M. & Ermert, H. (2005) Development and evaluation of a high-frequency ultrasound-based system for in vivo strain imaging of the skin. IEEE Trans. Ultrason. Ferroelectr. Freq. Control, 52, 375–385. doi:10.1109/TUFFC.2005.1417260 Voyiadjis, G.Z. & Peters, R. (2010) Size effects in nanoindentation: an experimental and analytical study. Acta Mech., 211, 131–153. doi:10.1007/s00707-009-0222-z Wade, W. (1856) Hardness of metals. in Reports on experiments on the strength and other properties of metals for cannon with a description of the machines for testing metals, and of the classification of cannon in service, pp. 259–275 and 313–314, Henry Carey Baird, Philadelphia. Walia, S., Shah, C.M., Gutruf, P., Nili, H., Chowdhury, D.R., Withayachumnankul, W., Bhaskaran, M., et al. (2015) Flexible metasurfaces and metamaterials: A review of materials and fabrication processes at micro- and nano-scales. Appl. Phys. Rev. doi:10.1063/1.4913751 Walley, S.M. (2012) Historical origins of indentation hardness testing. Mater. Sci. Technol. (United Kingdom), Vol. 28, pp. 1028–1044. doi:10.1179/1743284711Y.0000000127 Wang, Shang & Larin, K. V. (2015) Optical coherence elastography for tissue characterization: A review. J.

161

References Biophotonics, 8, 279–302. doi:10.1002/jbio.201400108 Wang, Shunfeng, Wang, X., Draenert, F.G., Albert, O., Schröder, H.C., Mailänder, V., Mitov, G., et al. (2014) Bioactive and biodegradable silica biomaterial for bone regeneration. Bone, 67, 292–304. doi:10.1016/j.bone.2014.07.025 Wells, R.G. (2013) Tissue mechanics and fibrosis. Biochim. Biophys. Acta - Mol. Basis Dis., Vol. 1832, pp. 884– 890. doi:10.1016/j.bbadis.2013.02.007 Welzel, J. (2001) Optical coherence tomography in dermatology: A review. Ski. Res. Technol., 7, 1–9. doi:10.1034/j.1600-0846.2001.007001001.x Wrucke, A.J., Han, C.S. & Majumdar, P. (2013) Indentation size effect of multiple orders of magnitude in polydimethylsiloxane. J. Appl. Polym. Sci., 128, 258–264. doi:10.1002/app.38161 Yang, Y., Bagnaninchi, P.O., Ahearne, M., Wang, R.K. & Liu, K.K. (2007) A novel optical coherence tomography-based micro-indentation technique for mechanical characterization of hydrogels. J. R. Soc. Interface, 4, 1169–1173. doi:10.1098/rsif.2007.1044 Yao, W., Yoshida, K., Fernandez, M., Vink, J., Wapner, R.J., Ananth, C. V., Oyen, M.L., et al. (2014) Measuring the compressive viscoelastic mechanical properties of human cervical tissue using indentation. J. Mech. Behav. Biomed. Mater. doi:10.1016/j.jmbbm.2014.01.016 Zeltmann, S.E., Bharath Kumar, B.R., Doddamani, M. & Gupta, N. (2016) Prediction of strain rate sensitivity of high density polyethylene using integral transform of dynamic mechanical analysis data. Polymer (Guildf)., 101, 1–6. doi:10.1016/j.polymer.2016.08.053 Zeltmann, S.E., Prakash, K.A., Doddamani, M. & Gupta, N. (2017) Prediction of modulus at various strain rates from dynamic mechanical analysis data for polymer matrix composites. Compos. Part B Eng., 120, 27–34. doi:10.1016/j.compositesb.2017.03.062 Zheng, Y., Wang, L., Li, T. & Wang, Y. (2014) An OCT-based air suction-indentation probe for tissue elasticity measurement. Opt. Elastography Tissue Biomech. doi:10.1117/12.2049254 Zorgkaartnederland (Institution). (2020) Dermatoloog en Dermatologisch Centra. Patiëntenfederatie Ned. Retrieved March 3, 2020, from https://www.zorgkaartnederland.nl/dermatoloog

162