INTRODUCTION: Image integration is being used in ablation procedures. However, the success of this approach is dependent on the accuracy of the image integration process. This study aims to evaluate the in vivo accuracy and reliability of the integrated image. METHODS AND RESULTS: One hundred twenty-four patients undergoing radiofrequency (RF) ablation catheter ablation for atrial fibrillation (AF) were recruited for this study from three different centers. Cardiac computerized tomography (CT) was performed in all patients and a 3D image of the left atrium (LA) and pulmonary veins (PVs) was extracted for registration after segmentation using a software program (CartoMerge, Biosense Webster, Inc.). Different landmarks were selected for registration and compared. Surface registration was then done and the impact on integration and the landmarks was evaluated. The best landmark registration was achieved when the posterior points on the pulmonary veins were selected (5.6 +/- 3.2). Landmarks taken on the anterior wall, left atrial appendage (LAA) or the coronary sinus (CS) resulted in a larger registration error (9.1 +/- 2.5). The mean error for surface registration was 2.17 +/- 1.65. However, surface registration resulted in shifting of the initially registered landmark points leading to a larger error (from 5.6 +/- 3.2 to 9.2 +/- 2.1; 95% CI 4.2-3.05). CONCLUSION: Posterior wall landmarks at the PV-LA junction are the most accurate landmarks for image integration in respect to the target ablation area. The concurrent use of the present surface registration algorithm may result in shifting of the initial landmarks with loss of their initial correlation with the area of interest.

Department of Cardiovascular Medicine Cleveland Clinic Cleveland Ohio 44195 USA