Generalized blind delayed source separation model for online non-invasive twin-fetal sound separation: a phantom...

Area of Science:

• Biomedical Engineering
• Maternal-Fetal Medicine
• Signal Processing

Background:

• Fetal phonocardiograms (fPCGs) are acoustic recordings of fetal heart mechanical activity.
• fPCGs provide vital indicators of fetal cardiac function and well-being, including heart rate and phase durations.
• Monitoring individual fetuses in multiple pregnancies presents unique signal separation challenges.

Purpose of the Study:

• To develop and validate a non-invasive algorithm for estimating individual fetal phonocardiograms from mixed signals in multiple fetus pregnancies.
• To apply blind source separation techniques to isolate fPCGs from complex acoustic mixtures.
• To assess the algorithm's efficacy using both simulated and experimental twin pregnancy data.

Main Methods:

• Modeling the mixture of fetal phonocardiograms using a generalized pure delayed mixing model.
• Assuming mutual independence of individual fetal phonocardiograms.
• Applying blind source separation (BSS) techniques to extract source signals (fPCGs).
• Validating the algorithm through computer simulations and phantom experimental data simulating twin pregnancies.

Main Results:

• The proposed algorithm successfully estimated individual fetal phonocardiograms from simulated and experimental mixtures.
• Blind source separation effectively separated overlapping fPCG signals.
• The method demonstrated robustness in a twin pregnancy simulation scenario.

Conclusions:

• The developed algorithm offers a promising non-invasive approach for assessing individual fetal cardiac function in multiple pregnancies.
• This technique can enhance prenatal monitoring by providing distinct fPCG data for each fetus.
• Further research may extend this method to higher-order multiple gestations.