Recognition of Schizophrenia Patients by EEG Signal Using the STFT/CWT Layer with Group-Convolution

• Published in: 2023 RIVF International Conference on Computing and Communication Technologies (RIVF) pp. 378 - 383
• Main Authors: Huynh, Viet Quoc, Do Tran, Huy, Van Huynh, Tuan
• Format: Conference Proceeding
• Language: English
• Published: IEEE 23-12-2023
• Subjects: Brain modeling; Continuous Wavelet Transform; Continuous wavelet transforms; Convolution; Convolution neural network; Electroencephalography; Fourier transforms; Mental disorders; Short Fourier Transform; Training; Wavelet analysis
• ISSN: 2473-0130
• DOI: 10.1109/RIVF60135.2023.10471833

The classic method of diagnosing schizophrenia involves a qualified psychiatrist interviewing the patient. Furthermore, a severe psychological condition significantly harms individuals, making prompt and precise detection crucial. In this article, the research focuses on identifying the healthy controls or a schizophrenia person through the electroencephalography (EEG) signal. The convolution neural network was applied to construct the essential convolution combined and depth-wise separable convolution. Moreover, the Short Fourier Transform and Continuous Wavelet Transform layers were applied to create a suitable input for the convolution process in the network. The research results indicated that EEG signals effectively distinguish schizophrenia patients and healthy control subjects, with results of about 85.10%. Significant advancements have been contributed in comparison to the state of art techniques in this field.