Developing potential neuroimaging biomarkers for early diagnosis of various neurological disorders is now an active area of research. A fast and efficient diagnostic tool would help for early detection, monitoring clinical trials, and tracking the progression of a disease. In this talk, I will present a couple of recent research findings in this direction. In the first part, I will discuss a deep learning method for simultaneous classification and severity prediction of tinnitus disease, an auditory disorder which causes structural abnormalities in auditory regions of the brain. In particular, we integrate deep features from two modalities - T1-weighted and T2-weighted structural magnetic resonance imaging (MRI) data. In the second part, I will introduce a graph convolutional learning method for schizophrenia classification. In this work, we train the network using both structural connectivity graphs obtained from diffusion tensor imaging (DTI) data and functional connectivity from functional MRI data.