Abstract:Over 60% of epilepsy patients globally are children, whose early diagnosis and treatment are critical for their development and can substantially reduce the disease's burden on both families and society. Numerous algorithms for automated epilepsy detection from EEGs have been proposed. Yet, the occurrence of epileptic seizures during an EEG exam cannot always be guaranteed in clinical practice. Models that exclusively use seizure EEGs for detection risk artificially enhanced performance metrics. Therefore, there is a pressing need for a universally applicable model that can perform automatic epilepsy detection in a variety of complex real-world scenarios.

What problem does this paper attempt to address?

The paper aims to address the issue of automatic detection of epilepsy (EP), particularly in the early diagnosis and treatment of pediatric patients. The current challenges include: 1. **Inability to guarantee seizure occurrence in clinical practice**: In actual electroencephalogram (EEG) examinations, seizures are not always observed, which makes models that rely solely on ictal EEG signals potentially artificially enhanced in performance metrics. 2. **Limitations of existing methods**: Although machine learning-based methods perform well, they have limitations in terms of reliance on operator judgment for feature extraction, susceptibility to noise interference, and difficulty in handling inter-individual differences. To address these challenges, the research team proposed a new deep learning model—Temporal Convolutional Network with Self-Attention (TCN-SA). This model consists of two main components: the Temporal Convolutional Network (TCN) for extracting time-varying features from EEG signals, and the Self-Attention (SA) layer that assigns importance weights to these features to focus on critical information, thereby improving the classification accuracy of epilepsy detection. ### Main Contributions - **Model Design**: The TCN-SA model can effectively extract abnormal activity information from EEG signals and adjust learning weights through the self-attention mechanism to identify seizures. - **Experimental Validation**: The study used pediatric epilepsy datasets and the Bonn dataset for validation. The results showed that the model achieved significant performance in automatic epilepsy detection tasks, with accuracies of 95.50% and 97.37% (A vs. E) and 93.50% (B vs. E), outperforming several other deep learning architectures. - **Stability and Reliability Assessment**: Cross-validation further confirmed the stability and reliability of the TCN-SA model under different conditions. In summary, the TCN-SA model proposed in this study provides an efficient and highly adaptable tool for automatic epilepsy detection, especially suitable for complex real-world clinical scenarios, and holds significant application value.

Automatic detection of epilepsy from EEGs using a temporal convolutional network with a self-attention layer

Real-Time Epileptic Seizure Detection Based on Deep Learning.

Automatic Epileptic Seizures Joint Detection Algorithm Based on Improved Multi-Domain Feature of cEEG and Spike Feature of aEEG.

Time-Series Anomaly Detection Based on Dynamic Temporal Graph Convolutional Network for Epilepsy Diagnosis

An Epileptic Seizure Detection Method Based on TCN-LSTM

Epileptic Seizure Detection Based on Path Signature and Bi-LSTM Network With Attention Mechanism

Protecting the Future: Neonatal Seizure Detection with Spatial-Temporal Modeling

A Study on Seizure Detection of EEG Signals Represented in 2D

Early Prediction of Epileptic Seizure Based on the BNLSTM-CASA Model

Automatic detection of epileptic seizure based on one dimensional cascaded convolutional autoencoder with adaptive window-thresholding

Automated detection of abnormalities from an EEG recording of epilepsy patients with a compact convolutional neural network

A novel epilepsy detection approach using intrinsic multiscale entropy analysis and DSEAM-enhanced 1D-ResNets

Convolutional Transformer Networks for Epileptic Seizure Detection

End-to-end model for automatic seizure detection using supervised contrastive learning

EEG-Based Seizure detection using linear graph convolution network with focal loss

Detection Method of Epileptic Seizures Using a Neural Network Model Based on Multimodal Dual-Stream Networks

Automatic Seizure Detection and Prediction Based on Brain Connectivity Features and a CNNs Meet Transformers Classifier

Epileptic-Net: An Improved Epileptic Seizure Detection System Using Dense Convolutional Block with Attention Network from EEG

Automatic Diagnosis Of Epileptic Seizure In Electroencephalography Signals Using Nonlinear Dynamics Features

Seizure Detection Based on Lightweight Inverted Residual Attention Network

Automatic interictal epileptiform discharge (IED) detection based on convolutional neural network (CNN)