Utilizing fMRI and Deep Learning for the Detection of Major Depressive Disorder: A MobileNet V2 Approach

Depression and related mental health conditions pose substantial obstacles in both diagnosis and treatment. This study investigates the application of functional magnetic resonance imaging (fMRI) combined with deep learning (DL) techniques to develop a diagnostic tool for major depressive disorder (MDD). The research utilizes fMRI datasets and MobileNet V2 model. In terms of preparing data for DL model, a toolbox for Data Processing & Analysis for Brain Imaging (DPABI) was used for preprocessing of fMRI datasets and the outputs were converted to images. This study developed one model for fMRI using 2D images of each fMRI slice. The performance of this model was evaluated using metrics such as precision, recall, F1 Score and Mathew's correlation coefficient (MCC). The performance of the best DL model had F1-score of 97.7%, precision of 97.67%, recall of 97.74%, and MCC of 95.02%.