An error-bounded median filter for correcting ECG baseline wander

Area of Science:

• Biomedical Engineering
• Signal Processing

Background:

• Baseline wander (BLW) is a common electrocardiogram (ECG) artifact impacting wave pattern recognition.
• Existing BLW correction methods often require processing uncompressed ECG signals, limiting efficiency.
• Compressed ECG data offer advantages in storage and transmission, necessitating methods for processing in this format.

Purpose of the Study:

• To introduce a novel median filter, CM-Filter, designed for baseline wander correction.
• To enable efficient BLW correction directly on compressed ECG data.
• To develop a method that is effective for both slow and abrupt baseline wander.

Main Methods:

• Piecewise linear approximation (PLA) to represent ECG signals as synopses of straight lines within a maximum error bound.
• Development of the CM-Filter, a median filter operating on these synopses.
• Implementation of a "Quick-Finding" heuristic strategy within CM-Filter for quality-assured median value extraction.

Main Results:

• The proposed CM-Filter demonstrates high efficiency in execution time.
• The filter effectively corrects both slow and abrupt types of ECG baseline wander.
• The method allows for baseline correction on compressed ECG data, enhancing processing efficiency.

Conclusions:

• The CM-Filter provides an efficient and effective solution for ECG baseline wander correction.
• Processing BLW correction on compressed ECG data is feasible and advantageous.
• The novel approach offers a valuable tool for ECG signal processing and analysis.