Improving workflow in prostate MRI: AI-based decision-making on biparametric or multiparametric MRI

Abstract Objectives To develop and validate an artificial intelligence algorithm to decide on the necessity of dynamic contrast-enhanced sequences (DCE) in prostate MRI. Methods This study was approved by institutional review board requirement for study-specific informed consent waived. A convolutional neural network (CNN) developed 300 MRI examinations. Consensus two expert readers DCE acted as reference standard. The CNN validated a separate cohort 100 examinations from same vendor 31 different vendor. Sensitivity/specificity were calculated using ROC curve analysis results compared decisions made radiology technician. Results reached sensitivity 94.4% specificity 68.8% (AUC: 0.88) DCE, correctly assigning 44%/34% patients biparametric/multiparametric protocol. In 2% all patients, incorrectly decided omitting DCE. With technician reaching 63.9% 89.1%, use would allow increase 30.5%. achieved AUC 0.73 set Conclusions have assigned 78% biparametric or multiparametric protocol, with only requiring re-examination add sequences. Integrating this clinical routine could render on-table monitoring obsolete performing when needed.