CLASSIFICATION OF EEG SPECTROGRAM IMAGES WITH DEEP LEARNING MODELS FOR ALCOHOLISM DETECTION

Electroencephalogram (EEG) signals are time series that play an essential role in un- derstanding the electrical behavior of the brain. The complex structure of the brain makes the in- terpretation of EEG signals difficult. In this study, the classification of EEG signals based on image processing with deep learning is performed differently from traditional methods. Images of EEG signals obtained for the detection of alcoholism were used to classify healthy and alcohol-addicted individuals using a Convolutional Neural Network (CNN). Three models have been implemented in the experiments conducted on the EEG images: Resnet50, Xception, and custom CNN. The findings demonstrate that Xception achieves the best accuracy with 100% classification success.