Prostate diagnosis may benefit from some deep learning
AI finding a new niche in pathology.
By Nick Carne
The inexorable march of AI into more and more areas of medicine appears to have a new focus on pathology services.
Earlier this week Cosmos reported on an algorithm that US scientists say performs as well as human pathologists in classifying surgical samples from the 10 most common types of brain cancer.
Now Dutch researchers from Radboud University have developed a “deep learning” system they say is better than most pathologists at determining the aggressiveness of prostate cancer.
It actually does the very same job – analysing pieces of tissue (biopsies) taken from the prostate then calculating a Gleason score, which provides a grading system for assessing the aggressiveness of a cancer and the best treatment options.
What it adds, the researchers say, is consistency. The current approach is necessarily subjective, because whether and how a patient is treated may depend on the pathologist who assesses the tissue.
“The Gleason score is the strongest correlating predictor of recurrence for prostate cancer, but has substantial inter-observer variability, limiting its usefulness for individual patients,” Wouter Bulten and colleagues write in The Lancet Oncology. “Specialised urological pathologists have greater concordance; however, such expertise is not widely available.”
The system examined images of 5759 biopsies from more than 120 patients to learn what a healthy prostate is, and what more or less aggressive prostate cancer tissue looks like. It was developed to be able to delineate individual glands, assign Gleason growth patterns, and determine the biopsy-level grade.
"When we compared the performance of the algorithm with that of 15 pathologists from various countries and with differing levels of experience, our system performed better than 10 of them and was comparable to highly experienced pathologists," Bulten says.
It needn’t cut humans completely out of the loop, however. The researchers suggest it could “potentially contribute to prostate cancer diagnosis”.
“The system could potentially assist pathologists by screening biopsies, providing second opinions on grade group, and presenting quantitative measurements of volume percentages,” they write.