癫痫脑电信号自动识别方法研究  

Automatic Recognition and Quantification of Epileptiform Activity in the EEG

作　　者：余荆生[1] 周防汛 许克克[1] 兰从庆[1] 杨先国 蔡兰[1] 

机构地区：[1]中国科学院武汉物理研究所,武汉附二医院神经科

出　　处：《湖北医科大学学报》1996年第4期312-315,318,共5页

基　　金：湖北省科委重点项目

摘　　要：在目前临床脑电医生常用的脑癫痫信号识别方法基础上，提出了一种利用计算机自动识别脑癫痫信号的方法及实现过程，该方法通过对脑电信号的形态特征的提取与在时空域进行综合分析的方法，模拟脑电专家的逻辑推理过程，将脑癫痫信号分为５类：尖波、棘波、尖慢波、棘慢波和多棘慢波来识别。经３０例临床脑电信号检验，本系统检出率达９０．３８％，证明了系统的有效性。A new epileptiform activity automatic detection system is described based on the conventional method for analysis the EEG data. This system extracts features of EEG waves and separates epileptiform from nonepileptiform waves by making use of a wide spatial and temporal context. The epileptiform activity includes five types of waves: sharp wave, spike wave, sharp slow wave, spike slow wave and multispike slow wave.30 records of EEG were used for testing the system, and the detection ratio is 90. 38% which shows that the system is effective.

关 键 词：癫痫 脑电图 计算机 

分 类 号：R742.104[医药卫生—神经病学与精神病学] R741.044[医药卫生—临床医学]