Artificial Intelligence Compared to Radiologists for the Initial Diagnosis of Prostate Cancer on Magnetic Resonance Imaging: A Systematic Review and Recommendations for Future Studies

Abstract

Simple Summary Radiologists interpret prostate multiparametric magnetic resonance imaging (mpMRI) to identify abnormalities that may correspond to prostate cancer, whose status is later confirmed by MR-guided targeted biopsy. Artificial intelligence algorithms may improve the diagnostic accuracy achievable by radiologists alone, as well as alleviate pressures on the prostate cancer diagnostic pathway caused by rising case incidence and a shortage of specialist radiologists to read prostate mpMRI. In this review article, we considered studies that compared the diagnostic accuracy of radiologists, artificial intelligence algorithms, and where possible, a combination of the two. Our review found insufficient evidence to suggest the clinical deployment of artificial intelligence algorithms at present, due to flaws in study designs and biases caused by performance comparisons using small, predominantly single-center patient cohorts. Several recommendations are made to ensure future studies bear greater clinical impact. Abstract Computer-aided diagnosis (CAD) of prostate cancer on multiparametric magnetic resonance imaging (mpMRI), using artificial intelligence (AI), may reduce missed cancers and unnecessary biopsies, increase inter-observer agreement between radiologists, and alleviate pressures caused by rising case incidence and a shortage of specialist radiologists to read prostate mpMRI. However, well-designed evaluation studies are required to prove efficacy above current clinical practice. A systematic search of the MEDLINE, EMBASE, and arXiv electronic databases was conducted for studies that compared CAD for prostate cancer detection or classification on MRI against radiologist interpretation and a histopathological reference standard, in treatment-naïve men with a clinical suspicion of prostate cancer. Twenty-seven studies were included in the final analysis. Due to substantial heterogeneities in the included studies, a narrative synthesis is presented. Several studies reported superior diagnostic accuracy for CAD over radiologist interpretation on small, internal patient datasets, though this was not observed in the few studies that performed evaluation using external patient data. Our review found insufficient evidence to suggest the clinical deployment of artificial intelligence algorithms at present. Further work is needed to develop and enforce methodological standards, promote access to large diverse datasets, and conduct prospective evaluations before clinical adoption can be considered.