Analysis of ECG Signals by Dynamic Mode Decomposition

Area of Science:

• Cybernetics
• Biomedical Signal Processing
• Cardiology

Background:

• Cybernetic principles suggest system stability analysis can inform overall system properties.
• Stability analysis has not been applied to electrocardiogram (ECG) signals for disease diagnosis.
• Decomposing ECG signals into subsystems offers a novel approach to understanding cardiac function.

Purpose of the Study:

• To evaluate the efficacy of analyzing the stability of decomposed ECG subsystems for investigating overall ECG signal performance.
• To explore the potential of this method in aiding cardiac pathology diagnosis.
• To introduce Dynamic Mode Decomposition (DMD) as a tool for ECG signal analysis.

Main Methods:

• Utilized Dynamic Mode Decomposition (DMD) to decompose ECG signals into dynamic modes (DMs), representing subsystems.
• Applied DMD to ECG signals from seven cardiac pathologies: myocardial infarction, cardiomyopathy, bundle branch block, dysrhythmia, hypertrophy, myocarditis, and valvular heart disease.
• Extracted stability-related features from DMs and employed nine common classifiers for pathology classification.

Main Results:

• Features derived from DM stability were significant (p<0.05 after Bonferroni correction) in differentiating cardiac pathologies.
• The proposed DMD-based method demonstrated superior performance compared to all existing methods for cardiac pathology classification.
• The study successfully illustrated the method's applicability across diverse cardiac conditions.

Conclusions:

• Dynamic Mode Decomposition (DMD) offers a new spatial and temporal method for analyzing ECG signals.
• This approach provides novel insights into cardiac mechanisms, enhancing the understanding of underlying physiology and disease.
• The method holds significant potential for widespread application in future ECG signal analysis and diagnosis.