A real-time algorithm for tracking of foetal ECG sources obtained by block-on-line BSS techniques

The foetal ECG (FECG) can be digitally extracted in real-time from non-invasive recordings using Blind Source Separation (BSS) techniques. BSS suffers the permutation ambiguity, scrambling the estimated sources over time and then hampering the FECG visual and automated analysis. In this paper a block-on-line tracking algorithm, including an unsupervised morphological stage able of creating an average FECG beat, is presented. It allows the automatic identification of the FECG sources block-wise even in presence of permutations. The algorithm has been successfully tested on both real and synthetic signals, showing a percentage of correct foetal ECG peaks identification up to 93.44%. Its porting on the OMAP L137 embedded processor, with the OL-JADE FECG extraction algorithm, allowed the assessment of its real-time capabilities.