摘要
本文介绍了一个脑电癫痫波自动检测系统的粗筛部分 ,用基于自适应预测的方法把脑电信号的平稳和非平稳成分分离 ,之后在两种成分中分别筛选出相应的癫痫特征波 ,并特别重视了对慢波的提取。利用 8段典型的临床数据进行了测试 ,结果表明在误检率较低的情况下 ,粗筛系统对癫痫病理波的总检测率达 93 5 % ,实现了预定要求。
The preliminary screening stage of an automatic detection system for epileptiform waves in EEG is described in this paper. An adaptive prediction based nonlinear method is proposed to separate EEG signal into stationary and non stationary parts, so as to extract the basic components (spikes, sharp waves and slow waves) more effectively from these two parts. In our system, special attention is paid to the extraction of slow waves which are usually neglected in current literature. The proposed system has been evaluated with EEG records from 8 patients. A total of 93.5% epileptiform waves reported by experts were accurately detected with a comparatively low false detection rate.
出处
《中国生物医学工程学报》
CAS
CSCD
北大核心
2001年第2期97-103,共7页
Chinese Journal of Biomedical Engineering
基金
国家自然科学基金资助项目! (3 9670 2 12 )
关键词
脑电癫痫波
自动检测
自适应预测
平稳非平稳成分分离
慢波
癫痫
脑电图
Automatic epileptiform wave detection
Adaptive prediction
Stationary and non stationary components separation
Slow waves
