Search

The precise detection of epileptic seizure helps to prevent the serious consequences of seizures. As the electroencephalogram (EEG) reflects the brain activity of patients effectively, it has been widely used in epileptic seizure detection in the past decades. Recently, deep learning-based detection methods which automatically learn features from the EEG signals have attracted much attention. However, with deep learning-based detection methods, different input formats of EEG signals will lead to different detection performances. In this paper, we propose a deep learning-based epileptic seizure detection method with hybrid input formats of EEG signals, i.e., original EEG, Fourier transform of EEG, short-time Fourier transform of EEG, and wavelet transform of EEG. Convolutional neural networks (CNNs) are designed for extracting latent features from these inputs. A feature fusion mechanism is applied to integrate the learned features to generate a more stable syncretic feature for seizure detection. The experimental results show that our proposed hybrid method is effective to improve the seizure detection performance in few-shot scenarios., Competing Interests: The authors declare that they have no conflicts of interest., (Copyright © 2022 Yayan Pan et al.)

The purpose of epilepsy detection is to determine whether epilepsy has occurred by analysing the patient's electroencephalogram (EEG) signals. Compared to traditional methods, epilepsy detection methods based on deep learning have achieved significant improvements in detection accuracy. However, when the number of training samples is limited, the model's detection performance often significantly declines. To address this issue, here a sample enhancement method based on electroencephalogram signal channel swapping is proposed. This method generates new electroencephalogram samples by exchanging electroencephalogram sequences from different channels, thereby expanding the training set and improving epilepsy detection accuracy in few‐shot scenarios. Experiments using the Children's Hospital Boston and the Massachusetts Institute of Technology (CHB‐MIT) dataset show that for training sets with 100, 500, and 1000 samples, detection accuracy improves from 0.6797 to 0.7789, 0.6952 to 0.8210, and 0.7273 to 0.8517, respectively. Compared to the sliding window method, the proposed method demonstrates higher accuracy in extreme low sample sizes. Combining both methods can further enhances detection performance, showing an improvement of approximately 8% across various configurations. [ABSTRACT FROM AUTHOR]