Fetal Electrocardiogram Extraction Using Hybrid Improved Deflation and Independent Component Analysis

The performance of fetal heart during pregnancy is considered as an approach to evaluatefetus health. In recent years, a lot of investigations are performed on electrical signalsextracted from mothers’ bodies in order to separate fetus heart signal from noisymultichannel observations. In this paper, a new method of Fetal Electrocardiogram (FECG)extraction is proposed which simultaneously applies Periodic Component Analyze (πCA),Independent Component Analysis (ICA) and Extended Kalman Filter (EKF) on MaternalElectrocardiogram (MECG) to accurately extract FECG signal. The basic part of theproposed method is Deflation method which is improved in this paper by adding a KalmanFilter (KF) at the final step to obtain accurate FECG. The second contribution of this paperis applying ICA approach on the output signal of KF to completely eliminate the remainednoise parts of FECG. Results of the proposed method on several groups of simulated datasetsprove its superiority to the other FECG extraction methods in terms of Signal-to-InterferencePlus Noise Ratio (SINR) improvement.Index Terms— Deflation , EKF , FECG extraction, πCA, ICA.