Researchers at Uppsala University in Sweden have found that artificial intelligence (AI) can be used to identify subtle tissue changes showing prostate cancer long before it becomes visible to pathologists. This could help identify men who have a high risk of developing prostate cancer and who should be screened regularly.
Prostate cancer is the most common form of cancer in men and the third most common cancer in the world after lung and colorectal cancer. Within their lifetime, one in eight men will be diagnosed with prostate cancer, although the global incidence varies greatly. For example, men in Northern Europe and men of African descent have a much higher risk of developing prostate cancer than men of Asian descent.
The risk of developing prostate cancer increases with age. Due to widespread prostate-specific antigen (PSA) screening, around two-thirds of cancers are detected at an early stage, allowing for fast treatment. However, PSA tests often have low specificity, and the MRI screening and needle biopsies that follow can still miss prostate cancer even when an aggressive tumor is present.
Therefore, the researchers at Uppsala University wanted to find out if AI could help identify prostate cancer even before it becomes visible to the naked eye—in tissue biopsies and MRI scans.
In their study, published in the journal Scientific Reports under the title “Discovery of Tumor Indicating Morphological Changes in Benign Prostate Biopsies Through AI,” the researchers analyzed 213 patients with an average age of 65 who had raised serum PSA levels but whose initial prostate needle biopsy samples showed no signs of a tumor. Pathologists had therefore assessed these patients as healthy upon examining the biopsies.
Within 30 months (less than 2.5 years), 88 patients were diagnosed with prostate cancer, 57 of these with high-risk cancer, while the remaining patients were cancer-free eight years later.
“The study has been nicknamed the ‘missed study’, as the goal of finding the cancer was ‘missed’ by the pathologists,” explained lead author of the study, Carolina Wählby, PhD, professor of quantitative microscopy at the department of information technology and SciLifeLab, Uppsala University.
The research team trained an AI tool by teaching it to analyze each biopsy image, hoping that it would pick up subtle abnormal patterns in the biopsies of patients who later developed aggressive cancer. Then, the researchers tested the AI on an independent set of images.
“We have now shown that with the help of AI, it is possible to find signs of prostate cancer that were not observed by pathologists in more than 80 percent of samples from men who later developed cancer,” Wählby said.
“When we looked at the patterns that the AI ranked as informative, we saw changes in the tissue surrounding the glands in the prostate—changes also observed in other studies. This shows that AI analysis of routine biopsies can detect subtle signs indicating clinically significant prostate cancer before it becomes obvious to a pathologist.”
The research team hopes that their AI tool could be used as a complementary screening step in the future. It could help identify those people who are assessed as healthy, but who actually are at higher risk of developing prostate cancer, and who should therefore go to regular screenings to enable early detection and prevention of the disease.
