PC-3 human prostate cancer cells [Image courtesy of Adobe Stock]
The tool, known as Unfold AI, creates detailed three-dimensional maps of prostate tumors after analyzing MRI scans and biopsy data. It then offers physicians with precise visibility into tumor boundaries thus helping them accurately determine the true extent of cancer within the prostate.
Published in BJUI Compass and recently highlighted in UCLA news, the technique could reduce treatment failures by more than 70%, according to the researchers.
The study involved 204 men with localized prostate cancer who underwent partial gland cryoablation.
In the recent study, tumor volume was the strongest predictor of treatment success, outperforming traditional metrics like tumor grade. Patients with tumors smaller than 1.5 cubic centimeters showed significantly better outcomes.
Last year, Avenda Health published related findings in The Journal of Urology, showing that its AI tech could improve urologists’ and radiologists’ ability to identify cancer extent. That research found ‘AI-assisted cancer margins significantly outperformed cognitively-defined and hemi-gland margins,’ with a balanced accuracy of 84.7% compared to 67.2% and 75.9%, respectively. This assessment was conducted after physicians initially defined margins manually, followed by a minimum four-week washout period before using the AI software.
“By using AI to measure the size of a man’s prostate tumor more precisely, we can better predict who is likely to be cured with focal therapies like partial gland cryoablation,” said Dr. Wayne Brisbane, assistant professor of urology at the David Geffen School of Medicine at UCLA and first author of the study in a press release.
Partial gland cryoablation (or focal therapy) freezes and destroys only the diseased regions of the prostate, sparing surrounding healthy tissue. In theory, this approach minimizes side effects while preserving critical urinary and sexual functions. The main challenge, however, is ensuring the true extent of the cancer is properly identified and treated. Conventional imaging can underestimate tumor boundaries, which can lead to incomplete ablation and recurrence.
“Such a method has not been previously available. It’s important because tumor volume is a major determinant of treatment success or failure. Using AI to predict tumor volume and shape gives a clearer picture and could help choose better candidates for focal cryotherapy,” noted Dr. Leonard Marks, professor of urology at UCLA and senior author of the paper.
Unfold AI blends MRI data, pathology results, and PSA metrics in a single predictive engine. R&D teams in imaging, computer vision, or machine learning can glean best practices from how the authors validated their model. They correlated AI-predicted tumors with whole-mount prostate pathology in a separate radical prostatectomy cohort—demonstrating a strong correlation (R2 = 0.76) between AI-estimated volume and actual tumor volume.
Accurate tumor-volume estimation could become an eligibility tool for focal-therapy studies, helping trial designers more effectively stratify patients by predicted success.
Beyond simply indicating whether a patient is a candidate for focal therapy, three-dimensional tumor mapping can help surgeons plan the precise placement of cryoprobes or other ablative devices. This is a step toward more personalized, targeted surgical interventions.
The UCLA and Avenda Health team emphasizes the need for larger, multi-center trials to confirm results and evaluate long-term benefits. As they state in the BJUI Compass article, “A multicentre randomised trial to evaluate tumor volume in selecting focal therapy candidates is indicated.”
