Leveraging Automation for Adaptive Planning in MRI-Guided Prostate Cancer Focal Cryoablation

Motivation: Predicting freezing volume intraoperatively is essential for effective focal cryoablation of prostate cancer. The lack of a fast, accurate, and clinically viable method currently hinders the development of robust planning software for optimal cryo-needle placement. Goal(s): To develop an automated method to predict final iceball formation as cryo-needles are inserted. Approach: We use automated segmentation of needles and urethra from MRI images, which serves as input for a predictive model to estimate iceball formation. Results: Tested on five retrospective cases, the software demonstrated feasibility, enabling rapid, one-click predictions within seconds. Impact: Integrating automatic AI-based segmentation for rapid iceball prediction enables adaptive, real-time planning in cryoablation. This approach minimizes labor-intensive segmentation, allowing updates with each needle insertion to ensure comprehensive tumor ablation, enhancing precision, confidence, and efficiency in prostate cancer treatment.