Three-dimensional activation sequence imaging in a rabbit model

This paper evaluates a biophysical-model based three-dimensional (3-D) activation sequence imaging approach in a rabbit model. In this approach, cardiac electrical sources within the myocardial volume are represented by distributed equivalent current densities; a realistic heart-torso volume conductor model is built from the CT scans of the rabbit's torso; spatial-temporal regularization is applied when solving the inverse problem of current density estimation; and the activation time at every myocardial location is determined as the time point when the estimated local current density reaches its maximum amplitude. Computer simulations have been conducted to image the activation sequence initiated by pacing 11 sites throughout the ventricular myocardium. Under 20 mu V Gaussian white noise, the average correlation coefficient (CC) between the imaged and the simulated activation sequences is 0.92, the average relative error (RE) is 0.19, and the average localization error (LE) is 4.99mm averaged over 11 pacing sites. Even under 6 mu V Gaussian white noise, reasonable results can still be achieved by the present approach with CC = 0.89, RE = 0.22, and LE = 6.85mm. The simulation results demonstrate that the present 3-D imaging approach has reasonable accuracy and robustness against recording noises.