Fetal heart rrate processing based on adaptive least mean squared algorithm

This paper focuses on the design and verification of an adaptive system for processing signals obtained from two interferometric measurement sensors for fetal heart rate (fHR) monitoring. Measurement sensors are realized by Mach-Zehnder fiber optic interferometers placed on chest and abdomen of pregnant women. Based on real measurements, a simplified dynamic signal model of a distribution of heartbeat sounds (HS) through human body was realized. This model allows the design, realization, and verification of the adaptive system least mean squared (LMS) algorithm. This system enables to extract fetal heart rate from a sensor placed on the abdomen, where an unwanted maternal phonocardiogram (mPCG) impairs this important signal. Standard filtration methods cannot be used because phonocardiogram of fetus and mother operate on the same frequencies. The functionality of the designed system was verified by objective parameters such as Signal Noise to Ration (SNR), Sensitivity (S+), and Positive Predictive Value (PPV).