AttCNNnet: Attention Based CNN Network to Detect Seizures from EEG subjects

Abstract

Epileptic seizures impact about 1% of the global population, highlighting the need for effective diagnostic and monitoring tools. Traditional seizure detection methods often rely on manual analysis of EEG signals, which can be time-consuming and prone to human error. Recent advancements in artificial Intelligence, especially deep learning, have shown great promise in automating seizure detection. In this study, we propose an attention-based convolutional neural network (CNN) named as AttCNNnet for automated seizure detection using EEG spectrograms. The EEG signals were transformed into time-frequency spectrograms, which were then fed into a proposed model enhanced with an attention mechanism to improve feature extraction and classification accuracy. The developed approach determined improved accuracy and robustness compared to existing methods with an accuracy 99.33%. This improvement in accuracy can significantly advance real-time seizure detection, paving the way for clinical applications.