Blind Extraction of FECG Combining Periodicity and Kurtosis

Detection of the fetal electrocardiogram (FECG) from the maternal cutaneous electrode recordings is an important but hard task. To extract the FECG, two important blind source extraction algorithms have been proposed by Barros and Zhi-lin Zhang which using the periodicity and kurtosis, respectively. Barros´ algorithm is sensitive for the estimation error of the period of the FECG and Zhi-lin Zhang´s algorithm requires precise estimation of the kurtosis range of the FECG. In this work, a novel blind source extraction algorithm combing periodicity and kurtosis is proposed to conquer the two flaws of the BSE algorithm of Barros´ and Zhang´s. Experiments on real world FECG illustrate the validity and better performance of the proposed algorithm.