Recognition and Removal of Artifacts in EEG Recording with Adaptive Infomax Algorithm of Independent Component Analysis
Lei Gao · Hangtian yixue yu yixue gongcheng · 2008
Objective To find a new approach to recognize and remove the artifacts from different sources in EEG recording.Methods Infomax algorithm of independent component analysis and threshold set of nonlinear parameters were combined.The self-adaptive algorithm was firstly improved,and nineteen-channel Electroencephalograms(EEGs)which included electromyogram,eye-movement and some other artifacts,were decomposed with self-adaptive infomax algorithm.Then three parameters were calculated with nonlinear analysis for all the independent components,and artifacts could be identified automatically by the threshold settings.At last,after all the artifacts were removed,the rest components were projected to the scalp electrodes,and the clear EEGs can be obtained.Results It was showed that the various artifacts could be recognized and separated from the EEGs successfully with self-adaptive infomax algorithm on the basis of blind source separation technique,and removal of artifacts could be realized with signal reconstruction.Conclusion Self-adaptive infomax algorithm is a potential appoach to remove artifacts in physiological signal.