Diagnosis of epileptic seizures from electroencephalogram signals using log-Mel spectrogram and a deep learning CNN model

The traditional procedure of the electroencephalogram (EEG) examination is laborious, slow, and prone to many diagnostic mistakes, creating a need for automated systems capable of reliably detecting epileptic seizures or other significant neurological conditions. To address this issue, this study proposes a novel deep learning-based method for automating seizure detection using the log-Mel spectrogram of EEG signals. The method innovatively transforms EEG data into images by applying the logarithm to the Mel-filter bank spectrograms of the signals, allowing convolutional neural networks (CNNs) to classify the resulting image representations. A CNN model is designed and trained on these images for classifying EEG signals. The proposed method is evaluated on two publicly available datasets, for both binary and multi-class classifications. Experimental findings on the Bonn dataset show that the proposed method's accuracy, sensitivity, specificity, precision, and F1-score for five-class classification are 98.13 %, 95.33 %, 98.83 %, 95.59 %, and 95.32 %, respectively. For the three-class classification on the same dataset, the accuracy, sensitivity, specificity, precision, and F1-score are 99.60 %, 99.33 %, 99.67 %, 99.52 %, and 99.41 %, respectively. For the three-class classification on the NSC-ND dataset, accuracy, sensitivity, specificity, precision, and F1-score are 93.33 %, 90.00 %, 95.00 %, 91.89 %, and 89.50 %, respectively. The proposed approach outperforms existing seizure detection methods, demonstrating superior classification performance. To the best of our knowledge, it is the first to use log-Mel spectrograms of EEG signals for epileptic seizure classification. These findings highlight the effectiveness of combining log-Mel spectrograms with CNNs for EEG-based epilepsy seizure detection, showcasing its potential for real-world automated epilepsy diagnosis.