Learning to compress electrocardiogram signals on a quick response code.

Area of Science:

• Biomedical Engineering
• Digital Health
• Medical Informatics

Background:

• Electrocardiograms (ECGs) are crucial for diagnosing cardiovascular diseases (CVDs).
• Sharing ECG data in low and middle-income countries (LMICs) is challenging due to limited connected health (CH) infrastructure and privacy risks associated with paper records.
• Quick Response (QR) codes offer a potential solution for secure and efficient data transmission.

Purpose of the Study:

• To propose and validate a learning-based compression method for ECG data suitable for QR code embedding.
• To ensure the preservation of clinically essential information and patient privacy during ECG data transmission.
• To enable the integration of ECG data into connected health systems for improved CVD monitoring and early detection.

Main Methods:

• A learning-based compression technique was developed to preserve key clinical information from ECG signals.
• The compressed ECG data was losslessly encoded using the Brotli algorithm for QR code embedding.
• The method was validated using a public dataset of ECG recordings from healthy individuals and patients with 26 CVD pathologies.

Main Results:

• The proposed method achieved a Compression Factor (CF) of 82.37.
• For lead-II ECG signals, PRD was 2.70% and SSIM was 0.94.
• For lead-I ECG signals, PRD was 2.80% and SSIM was 0.94, demonstrating high data fidelity.
• The method outperformed existing state-of-the-art approaches in terms of compression and data preservation.

Conclusions:

• The developed learning-based compression and QR code embedding method offers a secure, efficient, and scalable solution for ECG data transmission.
• This approach facilitates the integration of ECG data into connected health systems, particularly in resource-limited settings.
• The method supports continuous monitoring and early detection of cardiovascular diseases, enhancing patient care.