Using contexts and R-R interval estimation in lossless ECG compression

Area of Science:

• Biomedical Engineering
• Signal Processing
• Data Compression

Background:

• Electrocardiogram (ECG) data requires significant storage and transmission bandwidth.
• Existing ECG compression algorithms often face limitations in achieving high compression ratios while maintaining lossless data integrity.
• Efficient compression is crucial for telemedicine, long-term monitoring, and large-scale clinical data analysis.

Purpose of the Study:

• To develop and evaluate a new lossless compression scheme specifically designed for ECG signals.
• To improve the efficiency of ECG data storage and transmission.
• To achieve superior compression performance compared to existing algorithms.

Main Methods:

• A novel lossless ECG compression scheme incorporating short-term prediction with context conditioning.
• Long-term prediction utilizing an R-R interval estimation algorithm.
• Investigation and selection of a low-complexity, reliable QRS detection algorithm.
• Coding of prediction residuals using Golomb-Rice (GR) codes, with GR-ESC escape codes for enhanced performance in specific contexts.

Main Results:

• The proposed lossless ECG compression scheme achieves significant data reduction, lowering storage needs from 12 to approximately 3-4 bits per sample.
• The algorithm demonstrates consistently superior performance compared to other waveform or general-purpose coding algorithms.
• The combination of short-term and long-term prediction, along with optimized coding, leads to effective compression.

Conclusions:

• The developed lossless ECG compression scheme offers a significant improvement in data storage efficiency.
• The algorithm provides a reliable and high-performance solution for ECG data compression.
• This method is well-suited for applications requiring efficient handling of large ECG datasets.