Enhanced epileptic seizure detection using CNNs with convolutional block attention and short-term memory networks

Analyzing the electroencephalography (EEG) signals of epilepsy patients can monitor the condition, detect and intervene in epileptic seizures in time. To enhance the lives of these patients, it is necessary to develop accurate methods to detect epileptic seizures. This study proposes a novel epileptic seizure detection method based on deep learning and attention mechanisms. This proposed method combines two deep learning models, Convolutional Neural Networks (CNN) and Long-Short-Term Memory Networks (LSTM), to automatically extract spatial and time series features that characterize epileptic seizures from EEG signals. Then, the convolutional block attention module (CBAM) is introduced to enable the deep learning model to focus on processing key information. Finally, parameter optimization and ablation experiments are performed on the CNN_CBAM_LSTM deep learning model composed of CNN, CBAM and LSTM on the publicly available Bonn University dataset, and the performance of epileptic seizure detection is evaluated. The CNN_CBAM_LSTM achieved an accuracy of 98.80 % in detecting three types of EEG signals from epilepsy patients. This model demonstrated superior performance compared to existing state-of-the-art methods. The CNN_CBAM_LSTM effectively detects epileptic seizures, offering significant improvements in the quality of life for epilepsy patients through early detection and intervention.