2 minute read
Artificial Intelligence and Cancer
We are constantly learning more about the potential of Artificial Intelligence (AI) to improve medicine and patient care. Generally speaking, AI is a computer program that applies an algorithm to a dataset to make decisions and predictions. To develop an algorithm, medical researchers create a set of rules or instructions enabling their computer to analyze volumes of historical data, make predictions, and discern abnormalities. This deep learning capability is becoming a tool for medical professionals to use in cancer detection, diagnosis, and treatment alternatives. A quick literature review will alert you to the importance of early cancer detection. The survival rate is significantly higher if the cancer (or merely the risk of cancer) is detected early. This improvement is where AI can make a real contribution to oncology. According to the CDC, cancer is the second leading cause of death after heart disease in the United States. Worldwide, lung cancer is the deadliest, resulting in 1.7 million deaths worldwide in 2020. Some lung cancers result from pulmonary metastasis, i.e., cancer that begins in another part of the body and spreads to the lung through the lymphatic system or bloodstream. Almost any cancer can metastasize to the lung, including breast, bladder, colon, kidney, and prostate.
by Dr. Chuck Cadle
In April, NBC News reported on a diagnostic procedure using low-dose computed tomography (LDCT) for lung cancer screening. The report highlighted a recent study where MIT researchers used AI to screen for lung cancer. The study, published in the Journal of Oncology, demonstrated that a deep learning model assessing the entire volumetric LDCT data could be developed to predict individual risk without requiring additional demographic or clinical data. The procedure, named Sybil, uses a computer linked to an X-ray machine that gives off a very low dose of radiation to make a series of detailed pictures of areas inside the body. The images are taken from different angles to create 3-D views of tissues and organs. These pictures were then compared to a 15,000-participant dataset of lung cancer pictures resulting in 6,282 LDCTs in the test set. For testing, Sybil’s input consisted of LDCT images only, i.e., no image annotation or clinical information was included. The program accurately predicted an individual’s future lung cancer risk from a single LDCT scan. Of course, this procedure needs further testing and improvement before wide acceptance, which is why the researchers published the computer code. To see the NBC News article, go to https://www.nbcnews.com/health/healthnews/promising-new-ai-can-detect-early-signs-lung-cancer-doctors-cant-see-rcna75982.
In March of last year, the National Cancer Institute published an article entitled “Can Artificial Intelligence Help See Cancer in New, and Better, Ways?” The conclusion was a resounding yes. For example, a scan generated by AI was compared to a scan by a radiologist with 15 years of experience. The AI program accurately identified prostate cancer. The radiologist, Dr. Ismail Baris Turkbey, explained that the [AI] model found the prostate and outlined cancer-suspicious areas without human supervision. He felt that AI would help less experienced radiologists find prostate cancer when it’s present and dismiss anything that may be mistaken for cancer.
Tests like mammograms and Pap tests regularly check patients for signs of cancer or precancerous cells that might become cancer. The treatment options can be more effective if physicians can diagnose and treat cancer early before it spreads — or even before it forms Scientists, oncologists, and radiologists are working together to develop AI tools that enhance screening tests for different types of cancer, including breast cancer. One group created an AI algorithm that can help determine how often someone should get screened for breast cancer. The model uses a person’s mammogram images to predict their risk of developing breast cancer in the next five years. The model was more accurate in various tests than the current tools used to predict breast cancer risk. NCI researchers have built and tested a deep-learning algorithm to identify cervical precancers that should be removed or treated.
The Leukemia and Lymphoma Society is another source of research. Leukemia is
4 4
NO NO NO
IF YOU CAN READ THIS please thank MEDICALEXAMiNER advertisers + PLEASE SUPPORT THEM. THEY MAKE THIS NEWSPAPER POSSIBLE.