Session S42.3

Wave Equation Based Interpolation on Volumetric Cardiac Electrical Potentials

DJ Swenson*, JG Stinstra, BM Burton, KK Aras, RS MacLeod

University of Utah
Salt Lake City, UT, USA

Measurements of multiple time signals (mapping) in electrophysiology depends heavily on interpolation to reconstruct electrical potential fields across the surface of the heart or intramurally through the myocardium. In order to study normal and abnormal propagation due to infarcts, ischemia, or other forms of conduction delay, it is necessary for interpolation techniques to preserve the steep potential gradients that define the wavefront location, even in the face of inadequate spatial sampling by the electrodes. The quality of interpolation is of special importance in studying the features of the border zone between healthy and abnormal tissue, a critical region in the genesis of reentrant arrhythmias. Previous research from our group has demonstrated the effectiveness of wave equation based (WEB) interpolation especially in preserving sharp gradients on cardiac surface mapping, but this approach has not been extended to volumetric data. We propose that WEB interpolation will improve the accuracy of volumetric interpolation of cardiac potentials and remove artifacts in the areas with sharp gradients compared to other commonly used approaches. For this implementation and evaluation, we used simulated extracellular potentials (kindly provided by Dr. Natalia Trayanova, Johns Hopkins University) computed from a high resolution model of an infarcted canine heart. Virtual intramural needle electrodes were placed in a small region of the simulated heart and the potentials mapped to the nearest electrode. The time signals at each needle electrode site then provided the input (measured) data for the interpolation methods while the original time signals provided the gold standard test data. We implemented and compared trilinear, volumetric Laplacian, WEB trilinear, and WEB volumetric Laplacian (a novel hybrid of WEB and Laplacian methods) interpolation schemes through an entire activation wave based on root mean square (rms) value calculated at each time step. The WEB interpolation showed a 5 % advantage in rms value over the other methods when applied to the leading and trailing edges of the wave front. However, the WEB methods performed about 10% worse then the others at time points that did not include sharp gradients. The trilinear interpolation performed better than the volumetric Laplacian interpolation for both non-WEB and WEB approaches. The trilinear interpolation showed some significant artifacts that were reduced by approximately 40% by the WEB method. WEB interpolation was shown to be distinctly better for targeted situations such as sharp gradients and to reduce artifacts. However, it did not outperform other interpolation methods under all conditions, such as during the slowly evolving phases. We conclude that all purpose solutions to interpolating potentials measured in the heart are limited, but that the volumetric WEB interpolation has advantages when evaluating sharp gradients or removing artifacts.

(Abstract Control Number: 127)