Approach to feature extraction of ERP / 医疗卫生装备

ABSTRACT

Feature extraction of event related potential (ERP) plays an important part in both fundamental and clinical researches for cerebral neurophysiology.Based on the prior knowledge of feature distribution of ERP,this paper introduces an approach to feature extraction of ERP from composite EEG,which is combined with wavelet multiresolution analysis (MRA) and radial basis function neural network (RBFNN).The components related to low-frequency response can be extracted from the wavelet decomposition coefficients by RBFNN.Then signal reconstruction is implemented to obtain the feature of ERP.Experimental result demonstrates that the approach is reliable.