Fetal electrocardiogram extraction by independent component analysis

• Main Authors: Panason Manorost, Nipon Theera-Umpon, Sansanee Auephanwiriyakul
• Format: Conference Proceeding
• Published: 2018
• Subjects: Chemical Engineering; Computer Science; Engineering; Mathematics
• Online Access: https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85050341837&origin=inward; http://cmuir.cmu.ac.th/jspui/handle/6653943832/58408
• Institution: Chiang Mai University

© 2017 IEEE. Electrocardiogram (ECG) is one of the most important health measurement methods to present the states of heartbeat as well as provide insights into the abnormalities of patients' health. In this work, we extract fetal ECG from the abdominal signals in Abdominal and Direct Fetal Electrocardiogram database by independent component analysis (ICA) to give the information about their health or state of heart. With ICA, maternal ECG and fetal ECG will be independently separated and the results are compared with fetal heart rate signal. Another distinctive feature of this paper is the coefficient function which is used for maternal ECG reduction. The weight function can reduce the maternal R-peaks and it is good for the abdominal signals where maternal R-peaks are dominant. The experimental results show that the extracted fetal signals have distinct fetal R-peaks. Although these signals are still interfered by EMG signals, their quality is rather impressive for fetal heart rate detection as the average of the accuracy can be achieved at 90.43%.