检测脑电癫痫的多头自注意力机制神经网络

doi:10.3778/j.issn.1673-9418.2104089

摘要/Abstract

摘要： 癫痫是一种危及生命且具有挑战性的神经系统疾病，目前基于脑电图（EEG）的癫痫检测方法依然存在很多挑战，脑电图信号是不稳定的，不同的病人表现出的癫痫发作模式不同，检测脑电信号耗时费力，不仅会给医务人员带来沉重的负担，还容易造成误检情况的发生。因此，研究高效的跨多患者的癫痫自动检测技术是非常有必要的。提出了一种基于多头自注意力机制神经网络的癫痫脑电检测方法（CABLNet），利用卷积层捕获脑电时序信号的短期时间模式和各通道之间的局部依赖关系，使用多头自注意力机制进一步捕获具有时序关系的短期时间模式特征向量的长距离依赖关系和时间动态相关性，将上下文表示送入双向长短时记忆网络（BiLSTM）提取前后方向的信息，用logsoftmax函数进行训练和分类。实验使用CHB-MIT头皮脑电数据库数据，灵敏度、特异性、准确率、F1-score分别为96.18%、97.04%、96.61%、96.59%，结果表明，提出的方法优于现有方法，在癫痫检测性能方面有显著提高，对癫痫的辅助诊断具有重要意义。

关键词: 脑电图, 癫痫, 癫痫检测, 深度学习, 多头自注意力机制

Abstract: Epilepsy is a life-threatening and challenging nervous system disease. There are still many challenges in the detection of epilepsy based on electroencephalogram (EEG). Because the EEG signal is unstable, different patients show different seizure patterns. In addition, EEG detection is time-consuming and laborious, which will not only bring heavy burden to medical staff, but also easily lead to false detection. Therefore, it is necessary to study an efficient automatic epilepsy detection technology across multiple patients. In this paper, an epileptic EEG detection method (convolutional attention bidirectional long short-term memory network, CABLNet) based on the multi-head self-attention mechanism neural network is proposed. Firstly, the convolution layer is used to capture short-term temporal patterns of EEG time series and local dependence among channels. Secondly, this paper uses the multi-head self-attention mechanism to capture the long-distance dependence and time dynamic correlation of the short-term time pattern feature vectors with temporal relationship. Thirdly, the context representation is sent into a bidirectional long short-term memory (BiLSTM) to extract the information in the front and back directions. Finally, logsoftmax function is used for training and classification. Using CHB-MIT scalp EEG database data, the sensitivity, specificity, accuracy and F1-score are 96.18%, 97.04%, 96.61% and 96.59% respectively. The results show that the proposed method is superior to the existing methods and significantly improved in epilepsy detection performance, which is of great significance to the auxiliary diagnosis of epilepsy.

Key words: electroencephalogram (EEG), epilepsy, seizure detection, deep learning, multi-head self-attention mechanism

仝航, 杨燕, 江永全. 检测脑电癫痫的多头自注意力机制神经网络[J]. 计算机科学与探索, 2023, 17(2): 442-452.

TONG Hang, YANG Yan, JIANG Yongquan. Multi-head Self-attention Neural Network for Detecting EEG Epilepsy[J]. Journal of Frontiers of Computer Science and Technology, 2023, 17(2): 442-452.

参考文献

[1] FISCON G, WEITSCHEK E, CIALINI A, et al. Combining EEG signal processing with supervised methods for Alzheimer’s patients classification[J]. BMC Medical Informatics and Decision Making, 2018, 18(1): 35.
[2] ZHOU Y, HUANG R, CHEN Z, et al. Application of approximate entropy on dynamic characteristics of epileptic absence seizure[J]. Neural Regeneration Research, 2012, 7(8): 572-577.
[3] GIGOLA S, ORTIZ F, D’ATTELLIS C E, et al. Prediction of epileptic seizures using accumulated energy in a multi-resolution framework[J]. Journal of Neuroscience Methods, 2004, 138(1): 107-111.
[4] BEHARA D S T, KUMAR A, SWAMI P, et al. Detection of epileptic seizure patterns in EEG through fragmented feature extraction[C]//Proceedings of the 2016 3rd International Conference on Computing for Sustainable Global Development, New Delhi, Mar 16-18, 2016. Piscatawa: IEEE, 2016: 2539-2542.
[5] BENGIO Y, COURVILLE A, VINCENT P. Representation learning: a review and new perspectives[J]. IEEE Tran-sactions on Pattern Analysis and Machine Intelligence, 2013, 35(8): 1798-1828.
[6] MEI Z N, ZHAO X, CHEN H Y, et al. Bio-signal complexity analysis in epileptic seizure monitoring: a topic review[J]. Sensors, 2018, 18(6): 1720.
[7] CRALEY J, JOHNSON E, JOUNY C, et al. Automated inter-patient seizure detection using multichannel convolutional and recurrent neural networks[J]. Biomedical Signal Processing and Control, 2021, 64: 102360.
[8] YANG J, HUANG X, WU H, et al. EEG-based emotion classification based on bidirectional long short-term memory network[J]. Procedia Computer Science, 2020, 174: 491-504.
[9] 贾子钰, 林友芳, 刘天航, 等. 基于多尺度特征提取与挤压激励模型的运动想象分类方法[J]. 计算机研究与发展, 2020, 57(12): 2481-2489.
JIA Z Y, LIN Y F, LIU T H, et al. Motor imagery classification based on multiscale feature extraction and squeeze-excitation model[J]. Journal of Computer Research and Development, 2020, 57(12): 2481-2489.
[10] 金欢欢, 尹海波, 何玲娜. 基于生成少数类技术的深度自动睡眠分期模型[J]. 计算机应用, 2018, 38(9): 2483-2488.
JIN H H, YIN H B, HE L N. Deep automatic sleep staging model using synthetic minority technique[J]. Journal of Computer Applications, 2018, 38(9): 2483-2488.
[11] LEA C, VIDAL R, REITER A, et al. Temporal convolutional networks: a unified approach to action segmentation[C]//LNCS 9915: Proceedings of the 14th ECCV Workshops on Computer Vision, Amsterdam, Oct 8-10, 15-16, 2016. Cham: Springer, 2016: 47-54.
[12] MORID M A, SHENG O R L, KAWAMOTO K, et al. Learning hidden patterns from patient multivariate time series data using convolutional neural networks: a case study of healthcare cost prediction[J]. Journal of Biomedical Informatics, 2020, 111: 103565.
[13] WANG K, LI K, ZHOU L, et al. Multiple convolutional neural networks for multivariate time series prediction[J]. Neurocomputing, 2019, 360: 107-119.
[14] YUAN Y, XUN G X, JIA K B, et al. A multi-view deep learning method for epileptic seizure detection using short-time Fourier transform[C]//Proceedings of the 8th ACM International Conference on Bioinformatics, Computational Biology, and Health Informatics, Boston, Aug 20-23, 2017.New York: ACM, 2017: 213-222.
[15] 卫作臣, 邹俊忠, 张见, 等. 基于贝叶斯最小风险的癫痫脑电自动检测算法[J]. 计算机应用研究, 2019, 36(12): 3729-3734.
WEI Z C, ZOU J Z, ZHANG J, et al. Automatic detection of epileptic EEG based on minimum Bayesian risk and rotation forest[J]. Application Research of Computers, 2019, 36(12): 3729-3734.
[16] SUBASI A. EEG signal classification using wavelet feature extraction and a mixture of expert model[J]. Expert Systems with Applications, 2007, 32(4): 1084-1093.
[17] RAGHU S, SRIRAAM N, TEMEL Y, et al. Performance evaluation of DWT based sigmoid entropy in time and frequency domains for automated detection of epileptic seizures using SVM classifier[J]. Computers in Biology and Medicine, 2019, 110: 127-143.
[18] LI P, KARMAKAR C, YEARWOOD J, et al. Detection of epileptic seizure based on entropy analysis of short-term EEG[J]. PLoS One, 2018, 13(3): e0193691.
[19] AMIN S, KAMBOH A M. A robust approach towards epileptic seizure detection[C]//Proceedings of the 2016 IEEE 26th International Workshop on Machine Learning for Signal Processing, Vietri sul Mare, Sep 13-16, 2016. Piscataway: IEEE, 2016: 1-6.
[20] WANG G, SUN Z, TAO R, et al. Epileptic seizure detection based on partial directed coherence analysis[J]. IEEE Journal of Biomedical and Health Informatics, 2016, 20(3): 873-879.
[21] TRUONG N D, NGUYEN A D, KUHLMANN L, et al. Convolutional neural networks for seizure prediction using intracranial and scalp electroencephalogram[J]. Neural Networks, 2018, 105: 104-111.
[22] ULLAH I, HUSSAIN M, QAZI E U H, et al. An automated system for epilepsy detection using EEG brain signals based on deep learning approach[J]. Expert Systems with Applications, 2018, 107: 61-71.
[23] VIDYARATNE L, GLANDON A, ALAM M, et al. Deep recurrent neural network for seizure detection[C]//Proceedings of the 2016 International Joint Conference on Neural Networks, Vancouver, Jul 24-29, 2016. Piscataway: IEEE, 2016: 1202-1207.
[24] ABBASI M U, RASHAD A, BASALAMAH A, et al. Detection of epilepsy seizures in neo-natal EEG using LSTM architecture[J]. IEEE Access, 2019, 7: 179074-179085.
[25] HU X M, YUAN S S, XU F Z, et al. Scalp EEG classification using deep Bi-LSTM network for seizure detection[J]. Computers in Biology and Medicine, 2020, 124: 103919.
[26] YAO X, LI X, YE Q, et al. A robust deep learning approach for automatic classification of seizures against non-seizures[J]. Biomedical Signal Processing and Control, 2021, 64: 102215.
[27] LI W, QI F, TANG M, et al. Bidirectional LSTM with self-attention mechanism and multi-channel features for sentiment classification[J]. Neurocomputing, 2020, 387: 63-77.
[28] GUO X, ZHOU H, SU J, et al. Chinese agricultural diseases and pests named entity recognition with multi-scale local context features and self-attention mechanism[J]. Computers and Electronics in Agriculture, 2020, 179: 105830.
[29] VASWANI A, SHAZEER N, PARMAR N, et al. Attention is all you need[C]//Proceedings of the 31st International Conference on Neural Information Processing Systems, Long Beach, Dec 4-9, 2017. Cambridge: MIT Press, 2017: 6000-6010.
[30] HOCHREITER S, SCHMIDHUBER J. Long short-term memory[J]. Neural Computation, 1997, 9(8): 1735-1780.
[31] YAO X, CHENG Q, ZHANG G Q. Automated classification of seizures against nonseizures: a deep learning approach[J]. arXiv:1906.02745, 2019.
[32] WEI Z, ZOU J, ZHANG J, et al. Automatic epileptic EEG detection using convolutional neural network with improvements in time-domain[J]. Biomedical Signal Processing and Control, 2019, 53: 101551.
[33] HUANG C B, CHEN W T, CAO G T. Automatic epileptic seizure detection via attention-based CNN-BiRNN[C]//Proceedings of the 2019 IEEE International Conference on Bioinformatics and Biomedicine, San Diego, Nov 18-21, 2019. Piscataway: IEEE, 2019: 660-663.