Toward Autonomous Prostate Cancer Clinical Significance Determination from Spectral/Statistics Features in Bi-Parametric MRI

Abstract

Simple Summary Deciding between active surveillance and treatment for prostate cancer patients often requires accurate assessment of prostate tumors. Independent fast quantitative analysis may provide a check and support for radiologists who conventionally visually inspect MRI and may follow protocols such as PI-RADS. Artificial intelligence is increasingly being applied to help clinical assessment of prostate tumors detected with multi-parametric MRI, just as in other fields. However, implementation of artificial intelligence and deep learning is computationally intensive, requires fast, high-end processing boards and, therefore, uses considerable energy and water resources for cooling. A less intensive, less wasteful, cheaper, and less environmentally destructive assessment approach is needed. The simpler spectral/statistics approach that mimics color vision was previously successfully applied in retrospective pilot studies of bi-parametric MRI of prostate cancer. The novel approach needs far fewer resources, is less computationally intensive, and is simpler than artificial intelligence to evaluate prostate tumors. However, these earlier spectral/statistics pilot studies required the intervention of an analyst and too much time for implementation in future large patient studies that are needed to validate the novel approach. This study developed, applied, and tested new automation tools to expedite the spectral/statistics approach. Automating the novel approach resulted in sufficiently high AUCs and a reduction in processing time, warranting future applications in large patient cohorts.