Diagnosis of Depression Based on New Features Extractive from the Frequency Space of the EEG

Major depressive disorder (MDD) is a prevalent psychotic disorder. Understanding the neurophysiological characteristics of MDD is very important in diagnostic and therapeutic applications. The use of electroencephalography (EEG) is being developed to identify brain activity mechanisms in MDD patients. The present study aimed to examine the efficacy of spectral features extracted from EEG signals based on reductionism and relativism methods in the diagnosis of MDD. Using the Welch analysis method, the best results were calculated with 6 optimal frequency features corresponding to 5 channels on 12-second segments. Using the K-Nearest Neighbor classifier, criteria such as accuracy, F1 score and Kappa were obtained as 86.79%, 85.71% and 0.73 respectively.