摘要
患者有头痛或抽搐等症状时 ,医生怀疑为癫痫 ,就要记录其脑电图 ,并分析其中有无癫痫样放电(EpileptiformDischarge ,ED)。尽管对ED的自动检测已取得了良好的结果 ,但无法避免因为伪差和类似ED的脑电波造成的假阳性。因此在全自动分析时 ,对于该类脑电图往往容易被误判 ,本来没有ED的脑电图被误判为含有ED。本文根据对 41份脑电图中用人工神经网络交叉识别网络输出峰值分布曲线的分析 ,建立一套全自动判断脑电图中是否有ED的方法 ,即使存在一定量的假阳性的情况下 ,仍能作出正确判断。
When a patient felt faint and was suspected to be an epileptic patient, EEG will be recorded and analyzed whether there were some ED in. Although good results were obtained in automatic detection, false positive detection resulted from artifacts and ED like waves cannot be avoided. So the EEG without ED might be misjudged. A method of automated determination presence or absence of ED in the EEG was established, by analyzing ANN output peak distribution curves of cross recognition of 41 recordings. This method can judge correctly despite some false positive recognition may exist in analyzing.
出处
《中国生物医学工程学报》
CAS
CSCD
北大核心
2001年第2期116-120,共5页
Chinese Journal of Biomedical Engineering
基金
上海市教委基金资助
