Automatic Diagnosis of Schizophrenia in EEG Signals Using Functional Connectivity Features and CNN-LSTM Model

Schizophrenia (SZ) is a mental disorder that threatens the health of many people around the world. People with schizophrenia always suffer from symptoms that include hallucinations and loss of coordination between thoughts and feelings. Using deep learning and connectivity characteristics, we present a method to detect SZ from electroencephalography (EEG) signals. In this study, the data set of the Institute of Psychiatry and Neurology in Warsaw, Poland has been selected and used for experiments. First, the EEG signals are divided into 25-second time frames during the preprocessing step. Then, in the feature extraction step, deep learning (DL) and functional connectivity features (FCF) are used simultaneously. The DL model includes a CNN-LSTM network, and the functional connectivity techniques include the synchronization likelihood (SL), Fuzzy SL (FSL), and simplified interval type-2 FSL (SIT2FLS) methods. In this step, the DL features and the each functional connectivity features are combined using a concatenate layer and finally classified by the sigmoid activation layer. To better evaluate the results, K-Fold with K = 5 was used in the classification step. The results show that the proposed method has been able to achieve 99.43% accuracy.