Differential diagnosis between dilated cardiomyopathy and ischemic cardiomyopathy based on variational mode...

Area of Science:

• Cardiology
• Biomedical Engineering
• Signal Processing

Background:

• Dilated cardiomyopathy (DCM) and ischemic cardiomyopathy (ICM) share clinical features, complicating diagnosis.
• Current diagnostic methods like coronary angiography are invasive, costly, and contraindicated for some patients.
• Previous ECG-based studies lacked interpretability and optimal classification performance due to mode-field connection neglect.

Purpose of the Study:

• To develop an advanced classification algorithm for differentiating DCM and ICM using ECG signals.
• To enhance the interpretability and diagnostic accuracy of modal decomposition techniques applied to cardiovascular disease.

Main Methods:

• Utilized Variational Mode Decomposition (VMD) to preprocess and decompose ECG signals into five intrinsic modes.
• Extracted bispectrums from VMD modes and combined them with frequency and nonlinear features to create a feature vector.
• Employed machine learning classifiers including Random Forest (RF), decision tree, support vector machine, and K-nearest neighbor for classification.

Main Results:

• The proposed VMD and high-order spectra method achieved superior classification performance for DCM and ICM.
• Achieved an accuracy of 98.21%, sensitivity of 98.22%, and specificity of 98.19% on a dataset of 75 subjects.
• Identified that mode 3 consistently provided the best performance among individual modes.

Conclusions:

• The developed algorithm significantly improves the automatic diagnosis of DCM and ICM from ECG signals.
• This non-invasive approach offers a promising tool to reduce diagnostic burden and healthcare costs.
• The findings highlight the potential of VMD and high-order spectra in advanced cardiovascular diagnostics.