Automatic digitization of paper electrocardiograms - A systematic review

Area of Science:

• Cardiology
• Biomedical Engineering
• Digital Health

Background:

• Digitizing paper electrocardiogram (ECG) records is vital for preserving and analyzing cardiac data.
• The process presents challenges including skew correction, binarization, and signal extraction.
• Existing methods, including deep learning, show promise but lack standardized evaluation.

Approach:

• This article provides a systematic review of current ECG paper record digitization methods.
• It examines various approaches to address digitization challenges.
• The review highlights the need for common databases and metrics for method comparison.

Key Points:

• Standardization of evaluation metrics and databases is lacking for ECG digitization methods.
• Limited accessibility of code and software impedes comparative analysis of different approaches.
• Deep learning and automation show potential but require further development and validation.

Conclusions:

• Improving code accessibility and establishing common evaluation standards are essential for advancing paper ECG digitization.
• Further research is needed to overcome existing challenges and facilitate reliable cardiac data analysis.
• Standardization will enable more robust comparison and adoption of novel digitization techniques.