Detecting schizophrenia using structural and synthetic cerebral blood volume MRI with deep learning

Currently discrimination of patients with schizophrenia from normal healthy subjects remains challenging. In this study, it is aimed to improve the performance of schizophrenia detection using structure T1-weighed scans and synthetic functional maps with one 3D Convolutional neural networks. The results show that our approach could outperform the current state-of-the-art method in terms of AUROC and discriminative brain regions related are also localized.