Automatic epileptic seizure detection model based on multi-channel recurrence plots and SE-VGG16
10.3969/j.issn.1005-202X.2025.11.014
- VernacularTitle:基于多通道递归图和SE-VGG16的癫痫自动检测模型
- Author:
Bo LI
1
;
Huiqi BAO
1
;
Ningning WEI
1
;
Mengmeng WANG
1
;
Weimin GAO
1
Author Information
1. 湖南医药学院医学信息与工程学院,湖南 怀化 418000
- Publication Type:Journal Article
- Keywords:
epilepsy;
triaxial accelerometer;
multi-channel recurrence plot;
squeeze-and-excitation module;
convolutional neural network
- From:
Chinese Journal of Medical Physics
2025;42(11):1494-1499
- CountryChina
- Language:Chinese
-
Abstract:
An automatic epileptic seizure detection model based on multi-channel recurrence plots and SE-VGG16 is proposed to enhance the accuracy of automatic detection of epileptic seizures.Firstly,triaxial accelerometer is used to collect patients'motion signals as a substitute of electroencephalogram signals,and these signals are converted into two-dimensional images through recurrence plots and then subjected to multi-channel fusion for generating multi-channel recurrence plots.Subsequently,squeeze-and-excitation module is employed to improve VGG16,which enhances the network's adaptability to different channels and enables precise classification.Experimental results show that the proposed model achieves the best performance across all indicators,attaining an accuracy of 99.3%,precision of 99.2%,recall rate of 98.9%,and F1-score of 99.0%.This model significantly boosts the accuracy and applicability of epileptic seizure detection,expands the potential of wearable devices for epileptic seizure monitoring in non-clinical settings,and provides a new technical pathway for epilepsy management and research.
