The Impact of Torso Signal Processing on Noninvasive Electrocardiographic Imaging Reconstructions

Area of Science:

• Biomedical Engineering
• Signal Processing
• Cardiovascular Research

Background:

• Noninvasive electrocardiographic imaging (NEI) aims to reconstruct epicardial potentials from torso surface measurements.
• Accurate reconstruction is crucial for understanding cardiac electrical activity, especially during pacing.
• Evaluating signal processing techniques is essential for optimizing NEI performance.

Purpose of the Study:

• To assess the efficacy of various signal processing methods for NEI of single-site pacing data.
• To compare high-frequency noise removal (HFR) and baseline drift removal (BDR) techniques.
• To determine the optimal application of these filters in the NEI pipeline.

Main Methods:

• Utilized torso-tank setups with Langendorff-perfused hearts (n=4) for simultaneous epicardial and torso potential recordings.
• Applied 49 signal processing methods, categorized into HFR, BDR, and combined HFR+BDR, to torso potentials.
• Solved the inverse problem and compared reconstructed electrograms and activation maps to directly recorded data.

Main Results:

• HFR alone did not significantly improve electrogram amplitudes or QRS waveform correlation.
• BDR and combined BDR+HFR significantly enhanced electrogram accuracy, correlation, and reduced errors (p < 0.05).
• While BDR+HFR improved activation time and pacing site detection, BDR alone showed higher localization errors.

Conclusions:

• BDR is critical for improving reconstructed electrogram morphology and amplitude in NEI.
• The simplest BDR method (resetting the isoelectric point) is effective.
• HFR primarily impacts post-processing for feature extraction (e.g., activation times) and should be applied post-reconstruction to prevent over-filtering.