Novel Tool for Complete Digitization of Paper Electrocardiography Data

Area of Science:

• Biomedical Engineering
• Medical Informatics

Background:

• Paper-based electrocardiography (ECG) records pose challenges for large-scale retrospective studies.
• Extracting prognostic value from historical ECG data requires efficient digitization methods.

Purpose of the Study:

• To develop and validate a Matlab-based tool for converting paper ECG charts into digital signals.
• To enable long-term retrospective studies of cardiac patients by leveraging existing paper ECG archives.

Main Methods:

• Grayscale thresholding for grid detection on ECG charts.
• Column-wise pixel scanning for ECG signal digitization.
• Template-based optical character recognition for extracting patient demographic data.

Main Results:

• High correlation (0.85-0.9) between digitized and original ECG signals.
• Insignificant differences in clinically important ECG parameters (QRST complex, intervals) post-digitization.
• Strong intra-observer (0.8-1.00) and inter-observer (0.85-0.9) correlations for extracted parameters.

Conclusions:

• The developed tool accurately preserves essential ECG signal features, including the QRST complex.
• This digitization technique is suitable for retrospective analysis of large paper ECG databases.
• The tool facilitates integration of digitized ECG data with electronic medical records and digital analysis programs.