(iv) We further evaluated the TrainSet with respect to these metrics which are presented in the 5th row. For fairness, we augmented real CS PCa data to 1942 using the data augmentation approach proposed in Yang et al. (2017a), which increases the data volume via TPS-based non-rigid transformations, for training a multimodal classifier. By comparing all rows, we observe that the Real Data achieves the high- est IS values and the lowest FID value among all synthesis methods, implying that there still exists room for improvement in synthesizing truly realistic and varied mp-MRI data. However, the comparison results of SCA are encouraging. The classifier trained with the synthetic data from “Ours w/ the AD Maximization” achieves a slightly better performance than that with real ones, implying that our method could synthesize data with idepth features and is a more viable alternative for addressing the insufficiency of medical data than the traditional data augmentation for specific clinical tasks.