An Approach to Compute Fetal Cardiac Biomarkers from the Abdominal Electrocardiogram

Abstract

Objective: The fetal electrocardiogram (FECG) can be recorded from the ${20}^{th}$ week of gestation. The aim of this work is to determine fetal cardiac biomarkers from non-invasive cardiac signals that may be useful in the assessment of fetal health. Methods: We have developed an algorithm to obtain FECG fiducial points. It started by discriminating fetal heartbeats based on the relative location between fetal and maternal QRS complexes. An average beat is derived from the abdominal electrocardiogram (AECG) using 20 beats with a correlation greater than 0.95 and stable RR-interval, based on data from 12 fetuses (${38}^{th}$ - ${42}^{nd}$ weeks). We have implemented a combination between quaternion algebra and principal component analysis (Q-PCA method) to determine the onset and end of FECG waves by analyzing the angular velocity of the heart electrical vector. To validate our findings, we compared them with measurements obtained from the direct fetal electrocardiogram (DFECG), as a benchmark. Results: The values calculated by the Q-PCA method, as well as their correlation and the p-value in relation to the DFECG, were as follows: PR interval: $125.1\pm 19.8$ ms ($\rho =0.97$, p$<2.39e-7$), QRS complex: $73.0\pm 4.4$ ms ($\rho =0.67$, p$<1.74e-2$), QT interval: $261.1\pm 28.5$ ms ($\rho =0.84$, p$<7.05e-4$) and QTc interval: $388.3\pm 35.9$ ms ($\rho =0.79$, p$<2.27e-3$). Conclusion: Given its importance and the measurement performance achieved, the methodology presented represents a significant potential tool for improving the diagnosis of fetal health.