EEG Coherence as a Neuro-marker for Diagnosis of Schizophrenia

In this experimental study, an EEG coherence based approach is proposed for diagnosis of schizophrenia (sch). In this sense, coherence values estimated from 6 interhemispheric, 3 of left and right intra-hemispheric electrode pairs selected from 16 EEG channel system were used as feature vectors. Classification algorithms of k-nearest neighbor (k-NN), support vector machine (SVM) and multi-layer perceptron (MLP) were utilized for discrimination of coherences belonging sch and healthy (norm) participants. In proposed study, coherence measurements of sch patients were observed slightly lower according to norm groups over all brain regions. Increasing coherence measurements were observed at higher frequency bands (beta-gamma) for sch patients. While higher amplitude of coherence values are achieved for inter-hemispheric electrode pairs (F3-F4, C3-C4), diagnostic ratio of sch is also concvincing as compare with intra-hemispheric electrodes. High achievement of inter-hemispheric electrode pairs stems from definite distance between two probes located on different hemisphere. Moreover, diagnosis of sch is performed effectively at right hemisphere compared to left. In binary classification of sch and norm, highest accuracy was obtained as 99.22% using k-NN algorithm. Proposed work is thought to generate effective solutions for diagnosis of sch disorder in clinical applications.