Research On The Key Methods Of Ballistocardiogram Artifact Removal

Simultaneous electroencephalogram(EEG)and functional magnetic resonance imaging(fMRI)provide observation of brain activity with millimeter spatial resolution and millisecond temporal resolution,which is one of the most important technologies for studying human brain function.However,in simultaneous EEG-fMRI acquisition,the obtaind EEG is contaminated by gradient artifact(GA)and ballistocardiogram(BCG)artifact.The GA is usually removed by template method.The BCG artifact has the characteristics of low signal-to-noise ratio,time and frequency confusion,unpredictability and time-varying,so the conventional filtering methods and template methods cannot effectively remove the BCG artifact in the obtained EEG.In order to effectively remove the BCG artifact in obtained EEG,a new BCG artifact removal method based on adaptive comb filter(ACF)combined with improved J-peak detection is proposed.In the method,the ECG is firstly used as the reference signal to obtain R-peak positions of ECG utilizing R-peak detection algorithm.The time delay is calculated between the each channel EEG and the obtained ECG.Then the time delay and R-peak positions are used to determine detection intervals of the corresponding J-peak positions.Next the accurate positions of J-peak of each channel are obtained using interval peak detection.Finally the BCG artifact is removed by the ACF with the accurately estimated J-peak positions as the parameters of ACF.The BCG artifact removal method based on least squares support vector machine(LSSVM)is proposed.Firstly,the model on the BCG artifact removal is established based on LSSVM.Then the LSSVM with optimized parameters is used to fit the nonlinear transformation between the ECG and the BCG artifact.Finally the optimal estimation of BCG artifact is obtained,and the BCG artifact is removed in the obtained EEG in simultaneous EEG-fMRI.The experimental results show that the BCG artifact removal method utilizing ACF combined with improved J-peak detection and the BCG artifact removal method based on LSSVM can effectively remove the BCG artifact in the EEG obtained by simultaneous EEG-fMRI.The two methods are compared with the traditional BCG artifact removal methods: adaptive comb filtering(ACF),principal component analysis(PCA),independent component analysis(ICA),the combination of PCA and ICA(PCA+ICA).In visual result,normalized power spectral density,normalized power spectral ratio(INPS)and peak-to-peak ratio,the two proposed BCG artifact removal methods are better than these traditional BCG artifact removal methods.The experimental results verified the proposed BCG artifact removal methods.