脑电图信号多维度特性分析在癫痫病发作预测中的应用
Application ofMultidimensional Characteristic Analysis of Electroencephalogram Signal in Epileptic Seizure Prediction
努尔比亚·阿不拉江 1阿地力江·阿布力米提 1祖木来提·司马义 1阿不都米吉提·阿吉 1阿依夏·米吉提 1古丽乃则尔·麦麦提1
作者信息
- 1. 喀什地区第一人民医院神经内科,新疆喀什 844000
- 折叠
摘要
癫痫患者的非线性脑电信号存在规律难以分类识别等困境.本研究基于卷积神经网络结合多种智能寻优算法,构建联合式脑电信号分类模型,并通过实验验证其收敛性和分类性能.模型不同的频率对大脑的刺激下均能准确地测试脑电信号对应的变化规律,并选取数据集对其收敛效率进行测试,联合算法从第10次迭代的收敛速度明显优于其余算法,到200代时仍具备较大优势.联合算法比传统的极限学习机分类效率高出约10％.综合来看,该模型在实际的诊断场景下对癫痫患者的脑电信号起到采集剖析分类等作用,对癫痫发作的诊断和预测具备一定的实用性和参考价值.
Abstract
The nonlinear EEG signals of epilepsy patients face challenges such as difficulty in classifying and recog-nizing patterns.In view of this,this study constructs a joint EEG signal classification model based on convolutional neural networks combined with various intelligent optimization algorithms,and verifies its convergence and classifi-cation performance through experiments.The model can accurately test the corresponding changes in EEG signals under different frequencies of brain stimulation.And the convergence efficiency of the joint algorithm was tested by selecting a dataset.The convergence speed of the joint algorithm from the 10th iteration was significantly better than the other algorithms,and it still had a significant advantage in the 200th generation.The classification efficien-cy of the joint algorithm is about 10％higher than that of traditional extreme learning machines.Overall,this mod-el has played a role in collecting,analyzing,and classifying the EEG signals of epilepsy patients in practical diagnos-tic scenarios,and has certain practicality and reference value for the diagnosis and prediction of seizures.
关键词
癫痫/脑电信号/卷积神经网络/智能寻优算法/分类模型Key words
Epilepsy/Eeg signal/Convolutional neural network/Intelligent optimization algorithm/Classification model引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
万方数据