ABSTRACT
This chapter describes the computer-aided diagnosis (CAD) for prostate cancer detection using multiparametric magnetic resonance imaging and evaluates its application into clinical practice. The study population is composed of two different datasets: a training set used to develop and validate the CAD system, and a testing set on which the CAD performance was evaluated. The local Ethics Committee approved the study design, and all participants signed informed consent forms. Each prostate specimen was step-sectioned at 3 mm intervals perpendicular to the long axis of the gland. The first step to develop a CAD system which produces a voxelwise malignancy likelihood map of the prostate is image registration. The CAD system was first validated on the training set using the leave-one-out method, which involves training on all but one case, estimating the likelihood of malignancy of the voxels belonging to the left-out patient, and repeating the procedure until each case has been tested individually.
