DSNet: EEG-Based Spatial Convolutional Neural Network for Detecting Major Depressive Disorder

Major depressive disorder (MDD) is a mental disease that has a severe negative impact on people’s daily lives, which has become a leading global health burden. Previous neuroscience studies have proved that MDD patients have altered structural and functional connectivity between different brain regions compared to normal individuals. Measuring brain activities via electroencephalography (EEG) is a cost-effective and appropriate method for the detection of mental disorders such as depression. In addition, as deep learning (DL) is gaining attention in various research fields, increasing DL methods have been presented to diagnose depression. Inspired by these angles, this paper proposed an end-to-end spatial convolutional neural network (CNN) called DSNet for depression classification based on the resting-state EEG signals. Evaluated on a public dataset, our model obtained better classification performance with the accuracy of 91.69% via the leave-one-subject-out (LOSO) cross-validation strategy compared to other DL models. The experimental results demonstrate that DSNet can effectively extract information on spatial differences between depressed and normal individuals and could be a potential model for MDD detection.