Classification of Schizophrenia Patients by Nonlinear Analysis of EEG

In this study, the diagnosis of schizophrenia using the nonlinear analysis of electroencephalography (EEG) during a specific task has been considered. The task involves participants distinguishing between images containing unilateral and bilateral shapes. Due to the nonlinear nature of brain signals, the nonlinear analysis of EEG has recently received significant attention from researchers. Among the nonlinear measures, we used some types of entropy, correlation dimension, and some kinds of complexity. After applying preprocessing, these nonlinear metrics and the signal energy in the delta band, were used for feature extraction. In the next step, the data were applied to the SVM, K-nearest neighbors (KNN), and decision tree classifiers. In addition, we tried to use the capacity of simultaneous classifiers to reach better performance. For this purpose, an ensemble learning random forest classifier has been used. Finally, to evaluate the classification, electrodiagnostic medicine criteria such as sensitivity, specificity, accuracy, and precision have been used. The results of these criteria are equal to 94.03, 94.67, 94.06, 95.32 percent, respectively, and indicate the good performance of the classification system.