Compressive sensing of electrocardiogram signals by promoting sparsity on the second-order difference and by using...

Area of Science:

• Biomedical Signal Processing
• Machine Learning
• Algorithm Development

Background:

• Electrocardiogram (ECG) signal reconstruction is crucial for diagnosing cardiac conditions.
• Existing methods for ECG signal enhancement face limitations with temporally correlated signals.
• Improving the accuracy and performance of ECG signal reconstruction remains an active research area.

Purpose of the Study:

• To propose a new algorithm for reconstructing electrocardiogram (ECG) signals.
• To develop a dictionary learning algorithm to enhance ECG signal reconstruction performance.
• To evaluate the proposed algorithms against state-of-the-art methods.

Main Methods:

• Signal reconstruction based on minimizing the lp(2d) pseudo-norm using a sequential conjugate-gradient algorithm.
• Dictionary learning algorithm employing iterative signal reconstruction and dictionary update steps.
• Comparison with lp(1d)-regularized least-squares and Bayesian learning algorithms.

Main Results:

• The proposed lp(2d) pseudo-norm minimization algorithm demonstrates improved reconstruction performance for ECG signals.
• The dictionary learning algorithm further enhances reconstruction accuracy for specific signal classes.
• The novel approach outperforms existing state-of-the-art algorithms in simulations.

Conclusions:

• The developed signal reconstruction and dictionary learning algorithms offer superior performance for ECG signal processing.
• The lp(2d) pseudo-norm approach provides a robust method for enhancing ECG signal quality.
• The findings suggest potential for improved clinical applications of ECG analysis.