ICA在视觉诱发电位的少次提取与波形分析中的应用本文提出一种基于扩展的独立分量分析(ICA)算法的视觉诱发响应少次提取方法
经与目前临床通用的相干平均法比较,只经三次平均,在波形整体和P100潜伏期的提取上,效果显著,获得医师欢迎,很有进一步开发潜力
关键词:独立分量分析;少次提取;人工神经网络分类号:R318
19ICAINTHESINGLE-TRIALESTIMATIONANDANALYSISOFVEPHongBo,TangQingyu,YangFusheng(DepartmentofElectricalEngineering,TsinghuaUniversity,Beijing100084)PanYinfu,ChenKui,TeiYanmei(BeijingFriendshipHospital,Beijing100050)ABSTRACTAnovelmethodbasedontheExtendedInfomaxofICA(IndependentComponentAnalysis)wasproposedforsingle-trialestimationofmultichannelVisuallyEvokedPotential(VEP)
Itsencouragingresultswereillustratedbybothcomputersimulationandclinicaldataapplication
Thenumberoftrialsneededwasreducedtothree,buttheresultwasclearerthanthatobtainedby50timesconventionalcoherentaveraging
Byanalyzingthetimecourseandspatialpatternoftheindependentcomponents