Resting-State EEG Classification of Children and Adolescents Diagnosed With Major Depressive Disorder Using Convolutional Neural Networks

The aim of this research was to use Convolutional Neural Network (CNN) based on VGG16 model to identify biomarkers in Children and Adolescents with Major Depressive Disorder (MDD) from age-matched healthy young individuals. For this purpose, resting-state eyes-closed electroencephalography (EEG) was pre-processed, and analysis was performed based on frequency bands and Regions Of Interest (ROI). Overall, results achieved an accuracy of 0.875 and F1-Score of 0.638, and further revealed lower Delta activity (0.856 accuracy and 0.539 F1-Score), higher Theta activity (0.895 accuracy and 0.717 F1-Score), and higher Alpha activity (0.804 accuracy and 0.606 F1-Score) which were classified in MDD compared to healthy individuals. In this study, Beta and Gamma activities were not found as biomarkers for MMD with higher accuracy. Moreover, results showed that while in MDD group, Delta frequency bands were featured in left temporal, occipital, bilateral frontal, and central regions; Theta frequency bands were featured in left temporal and frontal, left occipital, and central regions. The Alpha frequency band was further featured in left frontal, central, left occipital, and left temporal regions.