Computationally efficient real-time interpolation algorithm for non-uniform sampled biosignals.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Medical Devices

Background:

• Electrocardiogram (ECG) signals are crucial for diagnosing cardiac conditions.
• Baseline wander in ECG signals can obscure important diagnostic features.
• Existing methods for baseline drift removal have limitations in computational efficiency and accuracy.

Purpose of the Study:

• To develop a computationally efficient interpolation method for removing baseline drift in ECG signals.
• To improve the accuracy of ECG baseline drift removal compared to existing techniques.
• To validate the proposed method on both synthetic and real-world ECG data.

Main Methods:

• A novel interpolation algorithm utilizing piecewise linear equations based on detected isoelectric baseline points per heartbeat.
• Segmentation of interpolation intervals to apply different linear equations.
• Testing with synthetic sinusoidal data across various frequencies (0.05–0.7 Hz).
• Validation using real-world baseline wander data from the MIT-BIH Noise Stress Database.

Main Results:

• The proposed algorithm demonstrated a mean root mean square (RMS) error of 0.9 μV per heartbeat on synthetic data.
• On real ECG data, the method achieved a mean RMS error of 10.9 μV per heartbeat.
• The algorithm showed comparable or superior performance to cubic spline and linear interpolation methods in terms of RMS error.

Conclusions:

• The novel interpolation method is computationally efficient and effective for ECG baseline drift removal.
• The algorithm provides accurate interpolation of non-uniform ECG samples.
• This technique offers a valuable advancement for improving the quality of ECG signal analysis.