Joint Improved Fast Independent Component Analysis and Singular Value Decomposition for Fetal Electrocardiogram...

Area of Science:

• Biomedical Engineering
• Signal Processing
• Cardiology

Background:

• Accurate fetal electrocardiogram (fECG) extraction is crucial for non-invasive prenatal monitoring.
• Existing methods face challenges in noise reduction and signal separation, especially in single-channel recordings.
• Maternal electrocardiogram (mECG) interference is a primary obstacle in obtaining clear fECG signals.

Purpose of the Study:

• To develop an improved single-channel fECG extraction method.
• To enhance the signal-to-noise ratio (SNR) and accuracy of fECG signals.
• To provide a robust method suitable for detecting maternal or fetal arrhythmias.

Main Methods:

• Combined improved FastICA algorithm with Singular Value Decomposition (SVD).
• Utilized an overrelaxation factor in FastICA for maternal ECG component estimation.
• Employed SVD for denoising preliminary fECG estimates and proposed an improved ECG signal reconstruction matrix for arrhythmia analysis.

Main Results:

• The proposed method significantly improved SNR by approximately 5 dB compared to standard FastICA.
• Achieved high fECG extraction accuracy, reaching 96.54% on the PhysioNet database.
• Demonstrated effectiveness on both synthetic and real abdominal signal datasets, including those with arrhythmias.

Conclusions:

• The combined FastICA and SVD approach offers superior fECG extraction performance.
• The method is effective in improving SNR and accuracy for non-invasive fECG monitoring.
• The technique shows promise for clinical applications, particularly in identifying fetal and maternal arrhythmias.