Manifold Approximating Graph Interpolation of Cardiac Local Activation Time

Area of Science:

• Computational cardiology
• Medical imaging and signal processing
• Electrophysiology

Background:

• Accurate local activation time (LAT) mapping is crucial for catheter ablation of cardiac arrhythmias.
• Current LAT mapping methods demand numerous observations for precise interpolation of sparse signals from intracardiac electrograms (EGMs).
• Existing performance metrics inadequately assess the quality of interpolated LAT maps.

Purpose of the Study:

• To develop a novel spatial interpolation method for LAT signals requiring fewer observations.
• To establish a realistic sub-sampling protocol for testing LAT interpolation algorithms.
• To introduce a new color-based metric for quantifying perceived differences in LAT maps.

Main Methods:

• A graph signal processing framework was employed to address the spatial interpolation challenge as a semi-supervised learning problem on a manifold.
• The proposed method offers a closed-form solution for LAT signal interpolation.
• A novel color difference equation, rooted in color theory, was developed to evaluate visual discrepancies in generated LAT maps.

Main Results:

• Evaluation on seven LAT maps from four patients demonstrated excellent accuracy with significantly fewer observations.
• The proposed method achieved an average of 6% lower error compared to state-of-the-art techniques using only 100 observations.
• The realistic sub-sampling protocol validated the robustness and efficiency of the interpolation approach.

Conclusions:

• Graph signal processing methods offer a significant improvement for local activation time mapping in cardiac ablation procedures.
• The developed technique can potentially reduce patient surgical time by minimizing the number of required LAT observations.
• The new color-based metric provides a more accurate assessment of interpolated LAT map quality.