Source separation from single-channel recordings by combining empirical-mode decomposition and independent component...

Area of Science:

• Biomedical Signal Processing
• Data Analysis
• Machine Learning

Background:

• Multichannel blind source separation is well-established using techniques like independent component analysis (ICA).
• Single-channel signal decomposition methods, such as single-channel ICA (SCICA) and wavelet-ICA (WICA), have limitations.
• Separating mixed sources in single-channel biomedical recordings remains a challenge.

Purpose of the Study:

• To introduce a new method for single-channel signal decomposition.
• To evaluate the proposed method's performance against existing single-channel techniques.
• To demonstrate the practical applicability of the new algorithm in real-world scenarios.

Main Methods:

• A novel approach combining empirical-mode decomposition (EMD) with independent component analysis (ICA) was developed.
• The proposed EMD-ICA method was compared with SCICA and WICA using simulations.
• Performance was assessed, particularly under varying noise-to-signal ratios (SNRs).

Main Results:

• The proposed EMD-ICA method demonstrated superior separation performance compared to SCICA and WICA.
• Outperformance was especially significant in simulations with high noise-to-signal ratios.
• The algorithm's effectiveness was validated through two real-life biomedical signal applications.

Conclusions:

• The EMD-ICA method provides an effective solution for single-channel signal decomposition.
• This technique offers improved source separation accuracy, especially in challenging noisy environments.
• The method shows promise for advancing biomedical signal analysis and interpretation.