Artificial Intelligence in Medicine: AI Advances in Diagnosing Prostate Cancer
“Artificial intelligence systems can be trained to detect and grade cancer in prostate needle biopsy samples at a level comparable to experts.” So says a Urology Today report on the uses of Artificial Intelligence (AI) in diagnosing prostate cancer (PCa).
An overview of progress in AI applications was presented by Dr. Peter Humphrey (Professor of Pathology, Yale University) during the 2023 meeting of the Society of Urologic Oncology (Nov. 28-Dec. 1, Washington, DC). Dr. Humphrey is an expert in the diagnosis and grading of PCa based on biopsy tissue.
AI computer programs can be trained to analyze tissue samples based on digitized images of glass slides that have been marked by humans to identify cancerous cells. The process is similar to the way a human brain is taught to recognized differences. Think of a 4-year-old watching “Sesame Street.” The screen shows four fruits, three oranges and one apple, while a song is playing, “Three of these things belong together, three of these things are kind of the same, but one of these things is not like the others. Come on, can you play this game?” The child correctly shouts “Apple!”
A trained program can review hundreds of slides much faster than a human pathologist, and flag those that are highly suspicious for PCa. Such efficiency whittles down the number of images for pathologist to review, a huge time-saver. The question is, how accurate is the program compared with the human’s evaluation?
Dr. Humphrey cited several studies that demonstrated AI diagnostic performance on a par with experts. These studies included AI programs trained to detect PCa, assign a Gleason grade, define disease risk level, predict genomic abnormalities, and even generate pathology reports.
For example, one study compared a Deep Learning (DL) system versus general pathologists in terms of the rate of agreement with prostate pathology specialists who had previously graded 498 tumor containing biopsy specimens. The rate of agreement, said Dr. Humphrey, “was significantly higher for the deep learning system (72%) than for general pathologists (58%).” In addition, for distinguishing non tumor from tumor in a set of 752 slides, the rate of agreement with the specialists was “94% for the deep learning system and similar at 95% for general pathologists.”
Finally, Dr. Humphrey noted areas of challenge to wide adoption of AI tools for diagnosing PCa:
- Conceptual – How would AI function in the clinical routine of busy practices? Can it safely be given autonomy?
- Technical – How will laboratory infrastructure need to adapt to include AI tools? Will pathologists learn how to use them responsibly?
- Ethical – Can AI tools correct diagnostic inequalities? Will they prove cost-effective in terms of healthcare expenditures?
The potential benefits of AI include alleviated pathologists’ workload, identification of high-risk cell structures (e.g. cribriform pattern), and quantifying the amount and grade of tumor activity. When incorporated into current lab analysis, it can function as an assistive and quality assurance tool. However, the task of digitizing slides poses a practical obstacle to broad adoption. According to Dr. Humphrey,
Only 5-20% of pathology laboratories and hospitals have whole slide scanners. These scanners are costly and require information technology (IT) infrastructure for image management and storage and personnel to operate and manage devices. Furthermore, there are economic issues pertaining to reimbursement.
That said, Dr. Humphrey foresees the day when 3D images will simply be directly digitally created and submitted for AI analysis, which will streamline accurate diagnostics.
Dr. Sperling has been an early researcher and adopter of AI tools that are based on MRI prostate scans. It is encouraging that the application of AI within the field of prostate pathology is growing, since it benefits urologists, oncologists and radiologists as we work together to vanquish prostate cancer. No doubt, Dr. Humphrey’s presentation informed and inspired the conference participants who attended his presentation.
NOTE: This content is solely for purposes of information and does not substitute for diagnostic or medical advice. Talk to your doctor if you are experiencing pelvic pain, or have any other health concerns or questions of a personal medical nature.
- Artificial Intelligence