Abstract: Based on fast independent component analysis (FICA),the study of feature extraction and detection about ventricular late potentials (VLP) was conducted.The channel expansion method was proposed to get 4 channel data from 2 channel data.At last,the VLP is separated completely from the normal electrocardiogram by using FICA to carry out bind source separation.The simulation result shows that independent component analysis is an effective means to achieve the feature detection of VLP.

Key words: ventricular late potentials, independent component analysis, bind source separation