Functional connectivity-based EEG features to assist the diagnosis of post-traumatic stress disorder patients

A resting-state EEG-based computer-aided diagnosis (CAD) system could complement the traditional diagnostic error for post-traumatic stress disorder (PTSD) patients. The aim of this study is to develop an EEG-based CAD system for diagnosis of PTSD patients. To this end, eyes-closed resting-state EEG data were recorded from 77 PTSD patients and 58 healthy controls. Two different types of functional connectivity-based features were extracted, i.e., phase locking values (PLVs) and network indices (strength, clustering coefficient, and path length), for both sensor- and source-level. The classification performances of each feature set were evaluated using a support vector machine with leave-one out cross-validation. The best classification performance was achieved when using source-level PLVs (accuracy - 70.37% and area under curve (AUC) - 0.85). In our future studies, we will attempt to enhance the performance of our proposed CAD system by using deep-learning algorithms.