Multi-Source Electrical Signals Based on Non-Invasive Fetal Monitoring: Recent and Future

Non-invasive abdominal electrical fetal monitoring is utilized to assess the fetal status. Comparing to widely used fetal monitoring techniques, it has more prominent performance in monitoring precision, operating complexity, and users’ safety as well as comfortability. However, challenges such as low signal quality and being limited to single source signal affect early diagnostic precision, thereby increasing the risk of adverse fetal events. This paper summarizes and compares recent methods proposed for fetal electrocardiogram (ECG) extraction and feature recognition, and suggests potential optimization directions. Literature review reveals that studies focused on improving the quality of fetal monitoring through optimizing or fusing signal processing algorithms based on traditional techniques or machine learning and deep learning models. Results indicate that correlation coefficients for fetal ECG extraction can be enhanced up to 87.6%, with 100% sensitivity in feature detection and 87.9% rate in uterine contraction detection, which represent that the methods proposed by studies have remarkably heightened signal quality and monitoring performance. The two research directions each have their advantages in improving accuracy, robustness, and computational complexity for fetal ECG and contraction signal extraction and feature recognition, thereby enhancing the interpretability and accessibility of fetal monitoring. And by perfecting their limitations, more reliable early diagnosis of fetal health will be achieved in the future, furthermore, mend adverse birth outcomes.