Grant: Project Grant

Prostate cancer is one of the most commonly diagnosed cancers in Canadian men. Many men with low-risk disease are managed with active surveillance (AS) to avoid overtreatment, but AS relies on repeated prostate biopsies to catch progression. Biopsies are invasive, can cause infection and bleeding, and the burden of repeat procedures can reduce adherence over time. We need a safer, patient-specific way to monitor men on AS that can reduce unnecessary biopsies while still detecting clinically significant progression early. Multiparametric MRI and micro-ultrasound (a high-resolution ultrasound used in real time) can highlight suspicious areas, but neither is accurate enough to replace biopsy. They are also usually interpreted separately, which can leave uncertainty when results disagree or when benign conditions mimic cancer. We propose PROPEL-AS (Prostate Personalized Active Surveillance using Pathology-aligned Multimodal Imaging), an artificial intelligence (AI) decision-support system that integrates MRI and micro-ultrasound information over time, together with PSA density and prior biopsy findings. A key innovation is pathology alignment, where the AI will be trained so its imaging outputs match patterns seen by pathologists in prostate tissue. Our objective is to identify a low-risk subgroup in whom biopsy intervals can be safely extended, while maintaining high sensitivity for detecting progression to clinically significant cancer. Using multi-centre datasets, we will develop and validate PROPEL-AS, integrate it into the micro-ultrasound workflow, and evaluate it prospectively in a two-centre silent mode study where usual care is unchanged while the system's predictions are tested. PROPEL-AS will be used as a second reader to support clinician decision making. This work is expected to reduce unnecessary biopsies and improve adherence to AS, therefore increase survival with higher quality of life.