Geoffrey Sonn, MD, on an AI risk mapping tool to predict seminal vesicle invasion

Unfold AI, an artificial intelligence (AI)-based risk mapping tool, was able to accurately predict seminal vesicle invasion (SVI) in patients with Grade group 2 or higher prostate cancer, according to data presented at the 2025 American Urological Association (AUA) Annual Meeting in Las Vegas, Nevada.¹

In an interview with Urology Times®, co-author Geoffrey A. Sonn, MD, shared his thoughts on the significance and implications of these findings.

He explained, “Seminal vesicle invasion is an important part of assessing prognosis of prostate cancer, as well as in helping with staging and planning treatment. [The] current methods available to us are suboptimal in their accuracy in finding seminal vesicle invasion.”

Sonn is an associate professor of urology at Stanford University School of Medicine in California.

Sonn also noted, “In reviewing the literature, MRI can under-detect seminal vesicle invasion between 40% to 45% of the time. Indeed, that was part of this study, assessing how well MRI was doing. Between the 2 cohorts, it was roughly 40% to 50% of the time that MRI was incorrect. This can have an impact on providing inaccurate staging, and therefore us providing patients with inaccurate information about their true prognosis. It can also have an impact in treatment planning and selection.”

The other main finding, Sonn reported, was the AI tool was shown to perform better than MRI in assessing for SVI.

REFERENCE

1. Brisbane WG, Priester A, Mota SM, et al. Prediction of seminal vesicle invasion using artificial intelligence prostate cancer risk mapping. J Urol. 2025;213(5S):e483. doi:10.1097/01.JU.0001109868.09667.98.20