A single-observer, experiential study was conducted to understand how predictive models of prostate cancer on multiparametric MRI can be used clinically, and to determine whether such models have the potential to improve observer performance. A radiologist experienced in prostate MRI was asked to interpret mpMRIs for 34 patients before and after viewing model-generated predictive maps. Results show that the radiologist generally had low confidence in the accuracy of the predictive maps. However, his performance was significantly improved in the cases where he judged the predictive maps to be helpful. A multi-reader iteration of the study is planned.
This abstract and the presentation materials are available to members only; a login is required.