Motivation: Prostate cancer is the most common male cancer in the U.S. MP-MRI is effective at aiding diagnosis, however, it may yield false positive results or miss small lesions. Goal(s): This study used MP-MRI intensities as input to machine learning models to create radio-pathomic maps of prostate cancer. Approach: This study analyzed 236 prostate cancer patients' pre-surgical MRI and histology data to develop tumor detection models. Images were processed, segmenting tissue features. Custom software co-registered MRI and histology, and models were trained to predict cancer. Results: Histological feature models performed within one standard deviation of ground truth, while classification models achieved ~80% prediction accuracy. Impact: This innovative approach uses radio-pathomic mapping for non-invasive prostate cancer detection, offering a quantitative alternative to PI-RADS scoring, enhanced cancer localization, and potentially improving diagnosis, grading, and treatment planning for prostate cancer patients.
Duenweg et al. (Tue,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: