It is important to not only identify prostate cancer (PCa) when it is present, but also to determine the aggressiveness of PCa. In this work, we developed a novel two-stage classification model for simultaneous detection of PCa on prostate MRI and localization of aggressive, high-grade PCa, using both quantitative MRI and radiomic features. The first-stage classifier was trained to detect cancer on a voxel-wise basis, and achieved an AUC of 0.818. The second-stage classifier was trained to predict the aggressiveness of candidate regions automatically derived from the voxel-wise predictions of the first stage, and achieved an AUC of 0.779.
This abstract and the presentation materials are available to members only; a login is required.