Deep Learning Model for Automated Prostate Cancer Detection Using Micro-Ultrasound

The study evaluates the use of the net deep learning model for automated prostate cancer (PCa) detection on micro-ultrasound (MUS) images, utilizing the Prostate Risk Identification Using Micro-ultrasound (PRIMUS) protocol. Conducted with 41 patients, the research found that the model achieved a sensitivity of 0.77 and specificity of 0.85 in identifying clinically significant PCa lesions (grade group ≥2). While the model demonstrated high accuracy, it also faced challenges with false positives due to imaging artifacts. Ongoing improvements aim to enhance margin overlap and external validation of the model.