Noninvasive fetal electrocardiogram extraction: blind separation versus adaptive noise cancellation

Area of Science:

• Biomedical Signal Processing
• Computational Biology
• Maternal-Fetal Medicine

Background:

• Extracting fetal electrocardiogram (FECG) signals from maternal abdominal recordings is challenging.
• Existing signal processing methods lack comprehensive comparison with advanced techniques like blind source separation (BSS).

Purpose of the Study:

• To compare the efficacy of a BSS technique against Widrow's multireference adaptive noise cancelling (ANC) for FECG extraction.
• To evaluate the performance of optimal Wiener-Hopf solutions within the ANC framework.
• To validate the BSS model for FECG analysis using real-world data.

Main Methods:

• Applied a BSS procedure utilizing higher-order statistics.
• Implemented Widrow's multireference adaptive noise cancelling approach, including optimal Wiener-Hopf solutions.
• Utilized real multichannel electrocardiogram (ECG) recordings from pregnant women.

Main Results:

• The BSS technique demonstrated more robust performance in FECG extraction compared to the adaptive noise cancelling method.
• Experimental results confirmed the effectiveness of the BSS approach on actual patient data.
• The study verified the validity of the BSS model for this biomedical application.

Conclusions:

• Blind source separation (BSS) is a highly effective method for fetal electrocardiogram (FECG) extraction.
• BSS outperforms traditional adaptive noise cancelling techniques for FECG signal retrieval.
• The findings support the use of BSS in clinical settings for non-invasive fetal monitoring.