Spatial enhancement of ECG using diagnostic similarity score based lead selective multi-scale linear model

Area of Science:

• Biomedical Engineering
• Medical Imaging
• Cardiology

Background:

• Improving spatial resolution in electrocardiograms (ECG) is crucial for accurate cardiac diagnostics.
• Existing methods for ECG lead reconstruction have limitations in preserving diagnostic fidelity.

Purpose of the Study:

• To introduce and evaluate a novel patient-specific approach for enhancing ECG spatial resolution.
• To transform a three-lead ECG into a standard twelve-lead ECG while maintaining diagnostic accuracy.

Main Methods:

• A patient-specific model utilizing wavelet domain inter-lead correlation for ECG enhancement.
• A lead selection algorithm based on a novel diagnostic similarity score to identify optimal predictor leads.
• Evaluation using standard closeness measures, diagnostic measures, repeatability, and diagnosability assessments.

Main Results:

• The proposed model successfully transforms three-lead ECG into twelve-lead ECG, enhancing spatial resolution.
• The patient-specific approach demonstrates superior preservation of diagnostic information compared to existing models.
• The diagnostic similarity score effectively computes closeness between original and enhanced leads.

Conclusions:

• The developed patient-specific ECG enhancement method offers improved spatial resolution and diagnostic information preservation.
• This approach holds significant potential for advancing non-invasive cardiac diagnostics.
• The lead selection algorithm contributes to personalized and accurate ECG reconstruction.