Deformation Imaging: basics to clinical applications Samir Saha, MD, PhD, FACC Sweden No No financial financial disclosure disclosure but but II am am thankful thankful to to Alan Alan Fraser, Fraser, UK, UK, and and Tom Tom Marwick, Marwick, Australia Australia for for using using some some of of their their slides. slides.
Edler: Wanted to use ultrasound to diagnose mitral stenosis Ultrasonic Cardiogram in Mitral Stenosis Inge Edler and Arne Gustafson ACTA Medica Scandinavica 1957
Slide courtesy: Dr. Judy Hung,MGH
Mitral Stenosis by inference: delayed motion of LA wall in diastole
Relationship of CAD, LVEF on Survival
Mock MB, Circulation 1982;66:562-567
Relationship of LVESV, LVEF and Survival
ď‚›
Aurigemma GP et al, Curr Probl Cardiol 1995;20:368
STEMI GUIDELINE 2013: Assessment of LV Function (AHA, ACC)
I IIaIIbIII LVEF should be measured in all patients with STEMI.
Yes sure but how?????
2D and standard Doppler echo in 2013: pulmonary arterial HTN (N=25) or pulm venous HTN (N=21). RHC to estimate PVR, PCWP
A curvilinear relationship between pulm vasc resistance and the ACT/TAPSE ratio
A simple ratio with a cut-off of approx 2 separates between pulm HTN with or without increased PVR 3.9±1.3
P=0.008
P<0.001 p=ns
TAPSE by M-Mode
P=0.008
ACT by PW Doppler
1.6±0.9
1.5±0.9
2.8±1.2
Despite a non-linear correlation PAcT/PASP ratio can serve as a non-invasive predictor of PVR with a cut off value of <2 identifying patients with vascular resistance > 3 WU and thus useful in differentiating patients with pulmonary hypertension due to increased PVR and PCWP.
Misclassification of LVEF EF
CMR<30% CMR<35% CMR<40%
Clinical EF 42.4%
21.2%
15.2%
2D EF
21.8%
21.8%
10.9%
3DE EF
14.5%
9.1%
5.4%
Aasha Gopal, ASE webcast
Problems with 2D,M-mode, standard Doppler echocardiography Too much geometric assumption Load dependence Image quality Test retest variability Interobserver agreement is not optimal Objective quantification of volume and functions is not always possible. Strain imaging has the potential to overcome these limitations
Deformation imaging by echocardiography
1-dimensional strain 2-dimensional strain 3-dimensional strain 4-dimensional strain Rotational deformation
Essentially 2 types of strain mappings: strain (%) and strain rate (1/s) Strain is the absolute deformation and is expressed as percentage deformation, measured at end systole Strain rate is the rate at which deformation takes place and is expressed as 1/s, measured at peak systole Normal strain is approx. 20% for the LV about 30% for the RV Normal strain rate in the LV => 1.0/s at rest There exists a general consensus that strain is a marker of volume ( like ejection fraction), while strain rate is a marker of contractility Applies for longitudinal, radial, circumferential motions
Lind, et al EJE, 2006
Timing!!! A young man with genetically verified ARVC
Timing!!!! No regional IVRT: 1 D
Regional IVRT 1 D
From: Echocardiographic Assessment of Myocardial Strain
J Am Coll Cardiol. 2011;58(14):1401-1413. doi:10.1016/j.jacc.2011.06.038
Figure Legend: Strain Imaging in a Normal Subject
Examples of tissue Doppler imaging velocity, strain rate, and strain curves for a cardiac cycle from a subject with normal cardiac Date of download: L = length; Copyright The American College of Cardiology. function. V = Švelocity. 12/3/2012
All rights reserved.
All All tissue tissue Doppler Doppler parameters parameters are are related related & & all all are are derived derived from from myocardial myocardial velocities velocities Velocity
∆
∫
Velocity gradient
Displacement
= Strain Rate
∫
∆
Tissue deformation = Strain (ε)
.. so quality depends on optimal velocity recordings
Physiological (echocardiographic) relevance of strain
Interrelationship between regional strain rate(left) and contractility index dp/dt (left) and strain% with ejection fraction for normal myocardium and changes induced by different inotropic and chronotropic challenges. G.R.Sutherland et al JASE 2004
3D strain vs. LVEF ( Inter-changeable?)
38 patients with heart failure
81 patients without
Heart failure confirmed by clinical, biochemical and echocardiographic methods
3D speckle tracking echo that generates simultaneous volume and mechanical data in the same beat
Manual endocardial tracking
Tracking contour moves along the endocardial border during the entire cardiac cycle
Samir Saha, Aasha Gopal, Rena Toole, Anatoli Kiotsekoglou, 2012
1-Dimensional strain High frame rate ( >100 hz), angle dependent
Quantitative stress echo ( MYDISE, MYDID)
Color Doppler interface on grey scale
Left bundle branch block
Excellent temporal resolution
Constriction, restriction
Problems with noise (low signal to noise ratio)
Pathologic vs. physiologic hypertrophy (Athletes vs. hypertensives)
Regional function
Marfan syndrome
Tethering and translation
Dyssynchrony
1-dimensional strain ( left) and strain rate ( right)
1-D strain imaging in Marfan syndrome
Saha, Kiotsekoglou, Govind et al, 2011
Sensitivity
Which modality of quantitative stress echo ? Comparison of ROC characteristic curves MYDISE
Voigt
AUC 92%
AUC 90%
1.0 .75 .50 .25
Cx p<0.001
0.0 0.0 .25 .50 .75 1.00
1 - Specificity
2D strain (Speckle tracking echocardiography) The Modality Angle independent, non Doppler modality Frame rate 60-100 Hz Tracks the entire myocardium and follows the natural acoustic markers frame by frame Provides global longitudinal strain in Bull´s eye plot Torsional mechanics in addition to deformation imaging Untwisting velocity in diastole
Clinical Application Validated by sonomicrometry Normative data available Prognostic value studied Subclinical disease detection
Cardiac mechanics Huge application in CRT post PROSPECT
Methodology
Quantification: Strain Rate Imaging 2D STRAIN Velocity is estimated as a shift of each object divided by time between successive frames (or multiplied by Frame Rate)--> 2D velocity vector: (Vx, Vy) = (dX, dY) * FR
TVI STRAIN RATE IMAGING
[1/s]
(or trace above baseline) = expansion
Y New location dY
0
V2 V1
Old location dX
BLUE
Ultrasound beams
GREEN
=
no deformation
RED
(or trace below baseline) = contraction
SR=[v(r) – v(r+∆r)]/∆r
Velocity
Displacement
Strain%
Strain rate
Automated Function Imaging (AFI): The cornerstone of 2D strain
Left bundle branch block with CAD Stress AFI
Rest AFI
3D Strain (3D Speckle Tracking Echocardiography Advantages: Denisa M, ESC 2011) Full 3D: No foreshortening, no “out of plane” motion, no angle dependency Full parametric display: Segmental values in Bull´s eye plots displayed throughout systole and diastole Comprehensive: All strain components; 3D strain, principal strain , area strain, twist Simultaneous: Single beat acquisition of all segments, single apical approach: poor parasternal image is not a limitation Time-effective: Single dataset provides volume and function (validated by MR) EF, ESV, Strain, Twist, Dyssynchrony index
Cheung, Nature Rev Cardiology, Nov 2012 Forces along perpendicular axes, â&#x20AC;&#x153;principal strainâ&#x20AC;? Forces acting parallel to LV: Shear strain
Cheung, Nov 2012
Clockwise rotation at the LV base
Counter clockwise rotation at the LV apex
Studies of LV torsion in the setting of STEMI
No STEMI, normal
STEMI, Anterior wall
Govind SC, Saha S, Echocardiography (Echocardiography 2010)
Comparative diagnostic value cardiac magnetic resonance and 3D speckle tracking echocardiography in heart failure Samir Saha, Aasha Gopla, Anatoli Kiotsekoglou, Rena Toole
Parametric imaging in real- time 3D!! 80 subjects with normal LV function vs. 40 subjects with HF (NYHA 1-IV)
ď&#x201A;&#x203A; 3D longitudinal S%
ď&#x201A;&#x203A; 3D circumferential S%
Advantages of 3D speckle tracking 1.Speckles are NOT lost in space, as in 2D strain 2.Generates simultaneous motion and volume data
Intra-observer variability: 3D principal tangential strain (PTS%)
Regression analysis for intra-observer variability: 3D PTS% -15
Measurement 2
-20
Adjusted R 2= 0.82 ; ANOVA P<0.001
-25 -30 -35 -40
y = -2.0740 + 0.9536 x
-45 -50 -45
-40
-35
-30
-25
-20
Measurement 1
PTS% showed excellent intra- observer agreement (95% limits of agreement, -4.9 to 3.6), with slight over estimation (0.7%).
Receiver operating characteristic curve
AUC for PTS% (0.85)
3D EF%
0.85
0.78 to 0.92 0.036
0.85
0.78 to 0.92 0.034
0.82
0.74 to 0.90 0.040
3D Global LS %
0.85
0.78 to 0.92 0.036
3D Global CS %
0.77
0.68 to 0.85 0.044
3D Global RS %
0.79
0.71 to 0.87 0.042
ESV ml/m2 3D PTS%
Key points of this transatlantic collaborative project
RESULTS 1. 3D principal tangential S â&#x20AC;˘ %,3D EF%, 3D ESVi all 1. provide similar AUCs 2. Combination of 3D principal tangential S% and 2. 3D EF% provide incremental diagnostic value (normal vs. reduced LV function) â&#x20AC;˘
CONCLUSION/COMMENTS 3D ESVi may remain as a stand alone marker of LV dysfunction? Use of 3D EF% alone may not be reliable, but addition of 3D PTS% may be as good as ESVi ( R=0.93 for both)
More data needed for routine clinical use Saha, Gopal, et al, ASE highlights 2012
CMR LVEF% vs. 3D speckle LVEF%; P= 0.0001
Measures of LV function
AUC
SE a
95% CI b
3D speckle EDV, ml 3D speckle ESV, ml 3D speckle LVEF% Global 3D Principal Tangential Strain% Twist, degree CMR LVEF%
0.760 0.857 0.859 0.833 0.602 0.996
0.0441 0.0344 0.0349 0.0390 0.0561 0.00386
0.673 to 0.833 0.781 to 0.915 0.783 to 0.916 0.753 to 0.895 0.508 to 0.691 0.962 to 1.000
a
DeLong et al., 1988. b Binomial exact
Wald chi-squared value
Incremental predictive value of CMRLVEF for elevated plasma NT-proBNP in heart failure
3D-EDV
3D STE-EF 3D STE-SDI CMR-EF CMR-EF BE
Comparison of r values -0.59 for CMR and -0.49 for 3D speckle LVEF
R= 0.3 : p<0.05
Bland Altman Plot 3D speckle-LVEF vs. CMR-LVEF
STAT Highlights : CMR and NTproBNP data included ROC revelaed highest AUC for CMR-LVEF at 0.996 This is significantly higher than 3D speckle LVEF (AUC 0.86; p = 0.0001) Univariate analysis showed significant association between NT-proBNP vs. 3D EDV, 3D speckle LVEF. and CMR LVEF; all p <0.05) Multiple regression with backward elimination showed CMR-LVEF to the singl emost strongest predictor of NT-proBNP in HF (Wald Chi-square = 11; adjusted R-squared= 0.23, P = 0.001) CMR LVEF and 3D speckle LVEF both had similarly significantly negative assocoation with (log) NT-proBNP ( -0.48 vs. - 0.59; p for Pearson´s r = ns) Global 3D principal tangential strain can be interchanged with LVEF by 3D speckle tracking but not by CMR There is slight underestimation of LVEF by 3D speckle compared with CMR (bias= -7.3%).
Differences between 3D strain and 4D strain 4D measures deformation over time ( the 4th dimension) 4D strain requires multi beat acquisition compared to single beat acquisition in case of 3D High frame rate is absolutely essential for 4D (ca 25 vol/s for a heart rate of 60, 40 vol/s for a heart rate of 100 bpm) Area strain
4 D strain in apical infarct Longitudinal strain
Circumferential strain
Area strain
Final Comments/Conclusions Conventional 2D and standard Doppler echocardiography is irreplaceable 1-D strain is virtually being eliminated 2D strain is ready for clinical application (CAD, HF, aortic stenosis , EVEN in subclinical disease, in the community setting) 3D strain is feasible: needs more data 4-D strain needs further validation
Thank you