ECG-derived respiration methods: adapted ICA and PCA.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Cardiology

Background:

• Respiration monitoring is crucial for diagnosing and managing various diseases.
• Ambulatory measurements are increasingly important, driving the need for indirect respiration sensing like ECG-derived respiration (EDR).
• Existing EDR methods using Principal Component Analysis (PCA) and Kernel PCA (KPCA) show promise but require improvement.

Purpose of the Study:

• To propose and evaluate novel algorithms for deriving more accurate EDR signals.
• To enhance the performance of existing EDR techniques through adaptation and signal processing.
• To compare the efficacy of the proposed methods against established EDR techniques.

Main Methods:

• Development of an adapted independent component analysis (AICA) algorithm for EDR signal extraction.
• Extension of linear PCA using best principal component selection, termed adapted PCA (APCA).
• Application of smoothing spline resampling and bandpass filtering to improve EDR signal quality for all methods.

Main Results:

• The proposed AICA and APCA methods demonstrated statistically significant improvements over existing EDR techniques.
• AICA achieved a correlation coefficient of 0.84 and coherence of 0.90 with reference respiration.
• APCA achieved a correlation coefficient of 0.82 and coherence of 0.91 with reference respiration, outperforming KPCA (correlation 0.76, coherence 0.85).

Conclusions:

• AICA and APCA represent significant advancements in EDR signal extraction from single-channel ECG.
• These improved EDR methods hold potential for enhanced non-invasive, ambulatory respiratory monitoring.
• The findings support the utility of these advanced signal processing techniques in clinical diagnostics.