Research On Combined Source And Abdominal Electrode-sourced Fetal Electrocardiogram Extraction Method

Various fetal diseases can be diagnosed in early stage by fetal electronic monitoring,thereby imporving the effectiveness of treatment and reducing the incidence rate and mortality of perinatal infants.At present,the most widely used electronic monitoring technology in clinic is cardiotography(CTG),which provides both fetal heart rate and maternal uterine activity information.But it is difficult to obtain reliable instantaneous fetal heart rate variability information using CTG.Fetal electrocardiogram(FECG)is a more promising fetal monitoring technology due that it non-invasively monitors the fetal health and provides high-precision fetal heart rate information in real time.Therefor it is of great theoretical significance and practical value to research on FECG extraction.The existing non-invasive fetal ECG acquisition methods can be divided into two categories that require both abdominal and chest electrodes(combined source(CS)methods)and require only abdominal electrodes(abdominal electrode-sourced(AES)methods).The AES method is more convenient for clinical application.In this thesis,the combined source(CS)FECG extraction method utilizing LightGBM is proposed.Firstly,the LightGBM model applied to FECG extraction is constructed.According to the characteristics of FECG,then the conventional LightGBM model is improved by adding a regularization term,the SNR of extracted FECG,to the objective function.The SNR is estimated using cross correlation.Using the maternal ECG(MECG)as reference signal,the nonlinear transform between MECG and MECG component in the abdominal signal is estimated using improved LightGBM.And the optimal estimation of MECG component is obtained by MECG undergoing the estimated nonlinear transform.A noisy version of FECG is obtained by suppressing the optimal estimation MECG component.At last,the noisy FECG is enhanced by use of extended Kalman filter(EKF).The real data are adopted to validate the proposed CS FECG extraction method.The experimental results show that the proposed CS FECG extraction method is better than other conventional CS FECG extraction methods both on visual results and objective assessment.In this thesis,the abdominal electrode-sourced(AES)FECG extraction method utilizing EKF combined with Fast ICA is proposed.Firstly,for AES signals containing only maternal abdominal signals,the raw maternal abdominal signals were preprocessed to suppress baseline drift,power frequency interference and pulse artifacts.Then the MECG and FECG are estimated from maternal abdominal signals by Fast ICA,and the estimated FECG contains residual MECG component and other noise.With the R-peak positions of MECG,which were detected from the estimated MECG,the optimal estimation of residual MECG component is obtained by EKF.The noisy FECG was extracted by suppressing the estimated residual MECG component.At last,the R-peak positions of FECG were detected from the extracted noisy FECG,and the clear FECG is obtained by EKF with it.The real data are adopted to validate the proposed FECG extraction method utilizing EKF combined with Fast ICA.The experimental results show that the proposed AES FECG extraction method is able to extract clear FECG and better than other conventional AES FECG extraction methods both on subjective visual results and objective assessment.