On a computer screen, two identical black and white images of murky shapes sit side by side. Ismail Baris Turkbey, M.D., a radiologist with 15 years of experience, has outlined an area on the left side where he believes the fuzzy shapes represent a creeping, growing prostate cancer. An artificial intelligence (AI) computer program has done the same thing on the other side of the screen, and the results are nearly identical.
The black-and-white image is an MRI scan of a man with prostate cancer, which the AI program has analyzed thousands of times.
"Without any human supervision, the [AI] model finds the prostate and outlines cancer-suspicious areas," Dr. Turkbey explains. His hope is that the AI will assist less experienced radiologists in detecting prostate cancer and dismissing anything that could be mistaken for cancer.
When it comes to the intersection of artificial intelligence and cancer research, this model is just the tip of the iceberg. While the potential applications appear to be limitless, much of the progress has been focused on tools for cancer imaging.
Doctors use imaging tests in a variety of ways, including detecting cancer in its early stages, determining the stage of a tumour, determining whether treatment is working, and monitoring whether cancer has returned after treatment.
Researchers have developed AI tools in recent years that have the potential to make cancer imaging faster, more accurate, and even more informative. And this has sparked a lot of interest.
Computer programs, or algorithms, that use data to make decisions or predictions are referred to as artificial intelligence. To create an algorithm, scientists may develop a set of rules or instructions for the computer to follow in order for it to analyze data and make a decision.
Dr. Turkbey and his colleagues, for example, used existing rules about how prostate cancer appears on an MRI scan. They then used thousands of MRI studies to train their algorithm, some from people known to have prostate cancer and some from people who did not.
Other artificial intelligence approaches, such as machine learning, teach the algorithm how to analyze and interpret data. As a result, machine learning algorithms may detect patterns that the human eye or brain cannot detect. The ability of these algorithms to learn and interpret data improves as they are exposed to more new data.
Deep learning, a type of machine learning, has also been used in cancer imaging applications by researchers. Deep learning algorithms classify information in ways similar to how the human brain does. Deep learning tools employ "artificial neural networks," which mimic how our brain cells receive, process, and respond to signals from the rest of our bodies.
Click here to get $500 google ad credit
AI can automate assessments and tasks that humans can currently do but take a long time," said Harvard Medical School's Hugo Aerts, Ph.D. "A radiologist simply needs to review what the AI has done—did it make the correct assessment?" after the AI provides a result. Dr. Aerts went on. He added that while automation is expected to save time and money, it must still be proven.
AI has also demonstrated the potential to improve cancer detection in people who are experiencing symptoms. Dr. Turkbey and his colleagues at the NCI's Center for Cancer Research, for example, developed an AI model that could help radiologists identify potentially aggressive prostate cancer on a relatively new type of prostate MRI scan called multiparametric MRI.
Although multiparametric MRI produces a more detailed image of the prostate than standard MRI, radiologists typically require years of practice to read these scans accurately, resulting in disagreements between radiologists viewing the same scan.
Several deep learning AI models for lung cancer have been developed to assist doctors in detecting lung cancer on CT scans. On CT scans, some noncancerous changes in the lungs resemble cancer, resulting in a high rate of false-positive test results that indicate a person has lung cancer when they do not.
AI may be able to distinguish lung cancer from noncancerous changes on CT scans, potentially reducing the number of false positives and sparing some people from unnecessary stress, follow-up tests, and procedures, according to experts.
Comments
Post a Comment