Sparse representation-based ECG signal enhancement and QRS detection

Area of Science:

• Biomedical Engineering
• Signal Processing
• Cardiology

Background:

• Electrocardiogram (ECG) signal enhancement and QRS complex detection are crucial for heart disease diagnosis.
• Traditional methods using fixed bases like Fourier or wavelet transforms have limitations.

Purpose of the Study:

• To propose a sparse representation-based algorithm for ECG signal enhancement and QRS complex detection.
• To accurately approximate ECG signals, remove noise and artifacts (e.g., baseline wandering), and detect QRS complexes.

Main Methods:

• Modeling ECG signals as a superposition of learned structures (atoms) plus noise.
• Utilizing learned atoms and their properties for signal approximation and noise reduction.
• Modifying atoms with high kurtosis values as an indication function for QRS complex detection.

Main Results:

• The proposed algorithm accurately approximates the original ECG signal.
• Effective removal of noise and artifacts, including baseline wandering.
• Successful detection and localization of QRS complexes using modified atoms.

Conclusions:

• The sparse representation-based algorithm offers a robust and effective approach for ECG enhancement and QRS detection.
• This method outperforms traditional fixed-basis techniques.
• The algorithm demonstrates efficacy on both simulated and real-life ECG data, advancing heart disease analysis.