基于多频带路径签名特征的癫痫脑电图信号分类方法

doi:10.12141/j.issn.1000-565X.230586

华南理工大学学报(自然科学版) ›› 2024, Vol. 52 ›› Issue (7): 9-18.doi: 10.12141/j.issn.1000-565X.230586

• 电子、通信与自动控制 • 上一篇下一篇

基于多频带路径签名特征的癫痫脑电图信号分类方法

郭礼华¹(), 杨辉¹, 吴倩仪², 茅海峰³

^1.华南理工大学电子与信息学院, 广东广州 510640
^2.广州医科大学附属第二医院神经内科, 广东广州 510260
^3.广州医科大学附属第二医院急诊科, 广东广州 510260

收稿日期:2023-09-18 出版日期:2024-07-25 发布日期:2023-12-27
作者简介:郭礼华（1978—），男，博士，副教授，主要从事医学信息处理和图像分类研究。
基金资助:
广东省基础与应用基础研究基金资助项目(2022A1515011549)

Epilepsy Electroencephalogram Signal Classification Method Based on Multi-Band Path Signature Features

GUO Lihua¹(), YANG Hui¹, WU Qianyi², MAO Haifeng³

^1.School of Electronic and Information Engineering，South China University of Technology，Guangzhou 510640，Guangdong，China
^2.Department of Neurology，the Second Affiliated Hospital of Guangzhou Medical University，Guangzhou 510260，Guangdong，China
^3.Emergency Department，the Second Affiliated Hospital of Guangzhou Medical University，Guangzhou 510260，Guangdong，China

Received:2023-09-18 Online:2024-07-25 Published:2023-12-27
About author:郭礼华（1978—），男，博士，副教授，主要从事医学信息处理和图像分类研究。
Supported by:
the Basic and Applied Basic Research Foundation of Guangdong Province(2022A1515011549)

摘要/Abstract

摘要：

基于脑电图（EEG）信号的癫痫自动检测对癫痫的临床诊断和治疗有很大的帮助。由于大部分癫痫识别算法忽略了EEG信号的时序关系，为此，文中提出了一种基于多频带路径签名特征的癫痫EEG信号分类方法。此方法首先将EEG信号分解成5个不同频段的频带信号，再通过路径签名算法进行特征提取，然后采用局部主成分分析去除特征相关性并进行特征融合，最后将融合特征送入集成分类器中进行预测分类。由于路径签名可以更深入地挖掘EEG信号的相关关系，结合局部主成分分析后，文中方法可以获取更有鉴别性的癫痫分类特征。分别在时长超过2 000 s癫痫发作片段的本地医院私有数据集和开源的CHB-MIT癫痫数据集上，选用10折交叉进行实验验证，结果表明：在私有数据集上，文中方法的平均分类准确率达到97.25%，比经典的基于经验模态分解（EMD）的方法提高了3.44个百分点，比最新的基于长短期记忆网络（LSTM）+卷积神经网络（CNN）的方法提高了1.35个百分点；在CHB-MIT数据集上，文中方法的平均分类准确率达到98.11%，比经典的基于EMD的方法提高了5.20个百分点，比最新的基于LSTM+CNN的方法提高了2.64个百分点；在两个数据集上文中方法的分类准确率均优于其他对比方法。

关键词: 脑电图分析, 癫痫发作分类, 路径签名, 信号分析

Abstract:

Automatic detection of epilepsy based on electroencephalogram (EEG) signals is greatly helpful for clinical diagnosis and treatment of epilepsy. Most epilepsy detection algorithms ignore the temporal relation of EEG signals, therefore, this paper proposed an epilepsy EEG signal classification method based on multi-band path signature features. Firstly, EEG signals were decomposed into five frequency bands. Secondly, features were extracted using the path signature (PS) algorithm. Thirdly, features were fused after local principal component analysis (LPCA) removed the feature’s correlation. Finally, an ensemble classifier was used to predict epilepsy. Since the path signature can dig into the correlation of EEG signals, combined with local principal component analysis, the method proposed in the paper can obtain more discriminative epilepsy classification features.The comparative experiments of 10-fold cross-validation were conducted to validate this method on two datasets, i.e., the private dataset from a local hospital with more than 2 000 seconds of segments and the CHB-MIT epilepsy dataset. The results show that the average classification accuracy of the method reached 97.25% on the private dataset, which is higher than those of the classical EMD (Empirical Mode Decomposition) method and the up-to-date LSTM (Long Short-term Memory Network) + CNN (Convolutional Neural Network) method by 3.44 and 1.35 percentage points respectively. Moreover, the proposed method can achieve an average classification accuracy of 98.11% on the CHB-MIT dataset, which is higher than those of the classical EMD method and the up-to-date LSTM+CNN method by 5.20 and 2.64 percentage points respectively, and this method achieves the best classification accuracy than other methods on both datasets.

Key words: electroencephalogram analysis, classification of epileptic seizure, path signature, signal analysis

中图分类号:

TP18

郭礼华, 杨辉, 吴倩仪等. 基于多频带路径签名特征的癫痫脑电图信号分类方法[J]. 华南理工大学学报(自然科学版), 2024, 52(7): 9-18.

GUO Lihua, YANG Hui, WU Qianyi, et al. Epilepsy Electroencephalogram Signal Classification Method Based on Multi-Band Path Signature Features[J]. Journal of South China University of Technology(Natural Science Edition), 2024, 52(7): 9-18.

图/表 7

图1

表1

表2

表3

图2

表4

表5

参考文献 25

1	YANG J， SAWAN M ．From seizure detection to smart and fully embedded seizure prediction engine：a review［J］．IEEE Transactions on Biomedical Circuits and Systems，2020，14（5）：1008-1023.
2	ABDENNADHER M， SAXENA A， PAVLOVA M K ．Evaluation and management of first-time seizure in adults ［J］．Semin Neurol，2021，41（5）：477-482.
3	LIU J M， LIU Z L， MENG F X ．The economic burden of epilepsy in a sample of people with epilepsy in China ［J］．Epilepsy Research，2013，103（2）：288-293.
4	HAROWITZ J， CRANDALL L， McGUONE D，et al ．Seizure-related deaths in children：the expanding spectrum［J］．Epilepsia，2021，62（3）：570-582.
5	KALEEM M， GURVE D， GUERGACHI A，et al ．Patient-specific seizure detection in long-term EEG using signal-derived empirical mode decomposition （EMD）-based dictionary approach［J］．IEEE Biomedical Signal Process Control，2018，15（5）：157-165.
6	YAO X H， CHENG Q， ZHANG G Q ．Automated classification of seizures against nonseizures：a deep learning approach［EB/OL］．（2019-06-05）［2023-09-14］．.
7	WEI Z C， ZOU J Z， ZHANG J，et al ．Automatic epileptic EEG detection using convolutional neural network with improvements in time domain［J］．IEEE Biomedical Signal Process Control，2019，53：101551/1-11.
8	HUANG C B， CHEN W T， CAO G T ．Automatic epileptic seizure detection via attention-based CNN-BiRNN［C］∥ Proceedings of 2019 IEEE International Conference on Bioinformatics and Biomedicine．San Diego：IEEE，2019：660-663.
9	DUAN L J， WANG Z Y， QIAO Y H，et al ．An automatic method for epileptic seizure detection based on deep metric learning［J］．IEEE Journal of Biomedical and Health Informatics，2021，26（5）：2147-2157.
10	YAO X H， LI X J， YE Q，et al ．A robust deep learning approach for automatic classification of seizures against non-seizures［J］．IEEE Biomedical Signal Processing and Control，2021，64：102215/1-20.
11	杨涌，秦小林，林小光，等．基于数据增强和深度学习的特定患者癫痫检测分析方法［J］．生物医学工程学杂志，2022，39（2）：293-300.
	YANG Yong， QIN Xiaolin， LIN Xiaoguang，et al ．Epilepsy detection and analysis method for specific patient based on data augmentation and deep learning［J］．Journal of Biomedical Engineering，2022，39（2）：293-300.
12	ZHAO S， YANG J， XU Y，et al ．Binary single-dimensional convolutional neural network for seizure prediction［EB/OL］．（2022-06-08）［2023-09-14］．.
13	GODOY R V， REIS T J， POLEGATO P H，et al ．EEG-based epileptic seizure prediction using temporal multi-channel transformers［EB/OL］．（2022-09-18）［2023-09-14］．.
14	BASTANY Z， ASKARI S， DUMONT G A，et al ．Association of cortical spreading depression and seizures in patients with medically intractable epilepsy［J］．Clinical Neurophysiology，2020，131（12）：2861-2874.
15	REDDY G， RAO R ．Automated identification system for seizure EEG signals using tunable-Q wavelet transform［J］．Engineering Science and Technology，an International Journal，2017，20（5）：1486-1493.
16	HUANG Z， YANG X， HUANG S，et al ．SECP-Net：SE-connection pyramid network for segmentation of organs at risk with nasopharyngeal carcinoma［J］．Bioengineering，2023，10（10）：1119/1-14.
17	GUO L， CHEN D， JIA K ．Knowledge transferred adaptive filter pruning for CNN compression and acceleration［J］．Science China Information Sciences，2022，65（12）：229101/1-2.
18	赵荣超，吴百礼，陈祝云，等．多尺度时空信息融合驱动的图神经网络故障诊断方法［J］．华南理工大学学报（自然科学版），2023，51（12）：42-52.
	ZHAO Rongchao， WU Baili， CHEN Zhuyun，et al ．Graph neural network for fault diagnosis with multi-scale time-spatial information fusion mechanism［J］．Journal of South China University of Technology （Natural Science Edition），2023，51（12）：42-52.
19	CHEVYREV I， KORMILITZIN A ．A primer on the signature method in machine learning［EB/OL］．（2022-06-29）［2023-09-14］．.
20	TANG Y， WU Q， MAO H，et al ．Epileptic seizure detection based on path signature and Bi-LSTM network with attention mechanism［J］．IEEE Transactions on Neural Systems and Rehabilitation Engineering，2024，32（1）：304-313.
21	ALUJA-BANET T， NONELL-TORRENT R ．Local principal components analysis［J］．Qüestiió，1991，15（3）：267-278.
22	WU X， ZHOU J L， HE J X ．Maximum likelihood estimation image denoising using non-local principle component analysis［J］．Acta Photonica Sinica，2011，40（12）：1827-1832.
23	KAMBHATLA N， LEEN T K ．Dimension reduction by local principal component analysis［J］．Neural Computation，1997，9（7）：1493-1516.
24	AKYOL K ．Stacking ensemble based deep neural networks modeling for effective epileptic seizure detection ［J］．Expert Systems with Applications，2020，148（15）：249-258.
25	MURSALIN M， ZHANG Y， CHEN Y，et al ．Automated epileptic seizure detection using improved correlation-based feature selection with random forest classifier［J］．Neurocomputing，2017，241（7）：204-214.

算法	k₁ × k₂	准确率/%
LPCA	3 × 3	97.25
	5 × 5	97.17
	7 × 7	96.91
	9 × 9	97.07
	11 × 11	97.01
全局PCA		95.33

L₁	准确率/%		准确率/%
3	94.39	7	97.25
4	94.97	8	97.12
5	95.64	9	97.04
6	96.76	10	96.81

路径签名阶次	准确率/%	敏感度/%	特异性/%	特征提取耗时/s
路径签名阶次	准确率/%	敏感度/%	特异性/%	单个样本	全部样本
1	53.44	31.89	74.99	0.013 6	64.137 6
2	97.25	96.24	98.26	0.384 5	1 813.302 0
3	97.31	96.27	98.35	2.472 9	11 662.196 4
4	95.83	95.57	96.09	19.175 4	90 431.186 4

方法	准确率/%		敏感度/%		特异性/%
方法	私有数据集	CHB-MIT数据集	私有数据集	CHB-MIT数据集	私有数据集	CHB-MIT数据集
EMD	93.81	92.91	92.34	94.27	95.28	91.55
RNN	91.36	88.69	87.69	88.80	95.03	88.60
CNN+MIDS	84.11	84.00	72.93	72.11	95.29	95.89
ACNN+BiRNN	95.27		92.58	93.94	97.96	92.88
Metric学习	94.58	86.68	91.42	79.64	97.74	93.71
BiLSTM	86.82	88.63	85.71	87.00	87.93	88.60
LSTM+CNN	95.90	95.47	94.07	93.89	97.73	96.48
BSDCNN	94.42		94.69	94.69	94.15
TMC-ViT	96.27	95.73	96.76	96.46	95.78	96.09
文中PS+LPCA+RF	97.25	98.11	96.24	97.57	98.26	98.65

方法	准确率/%		敏感度/%		特异性/%
方法	私有数据集	CHB-MIT数据集	私有数据集	CHB-MIT数据集	私有数据集	CHB-MIT数据集
LPCA+RF	93.44	92.25	91.89	91.34	94.99	93.36
PS+LPCA	96.12	95.46	95.27	94.31	96.97	96.61
PS+RF	94.03	93.95	91.97	91.63	96.09	96.27
PS+CNN	94.89	94.71	93.68	93.19	96.10	96.23
PS+LPCA+Stacking	96.51	96.90	95.83	97.03	97.19	96.77
文中PS+LPCA+RF	97.25	98.11	96.24	97.57	98.26	98.65

基于多频带路径签名特征的癫痫脑电图信号分类方法

Epilepsy Electroencephalogram Signal Classification Method Based on Multi-Band Path Signature Features

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 7

参考文献 25

相关文章 1

编辑推荐

Metrics

本文评价