Wavelet denoising as a post-processing enhancement method for non-invasive foetal electrocardiography.

Area of Science:

• Biomedical Signal Processing
• Maternal-Fetal Medicine
• Cardiology

Background:

• Non-invasive fetal electrocardiogram (fECG) detection is challenging due to noise.
• Wavelet denoising is a common technique to improve signal-to-noise ratio (SNR).

Purpose of the Study:

• To systematically evaluate conventional and unconventional wavelet denoising methods for fECG.
• To assess the effectiveness of these methods in improving fECG SNR and preserving signal morphology.

Main Methods:

• Compared stationary wavelet transform (SWT) and stationary wavelet packet transform (SWPT).
• Evaluated three thresholds: Minimax, Universal, and an adapted unconventional threshold.
• Analyzed decomposition depth and used synthetic and real fECG datasets.

Main Results:

• Wavelet denoising generally improved SNR.
• SWT methods showed superior morphology preservation and SNR improvement over SWPT.
• The adapted Han et al. threshold yielded the best quality improvement.

Conclusions:

• Careful selection of wavelet denoising parameters is crucial for fECG post-processing.
• Signal morphology can be altered by wavelet method parameterization.
• Finer sub-band decomposition (SWPT) did not enhance processing quality.