Schweitzer, Jeffrey

Background: Current mapping systems use simultaneous displays of overlapping 3D maps and pre-acquired CT images of the LA to “pseudo-register” data sets rather than specifically “register” each point of one mapping data set to the other. The appropriateness of the 2 data set match has not been validated. Since significant surface to surface mismatch errors may occur, this study was undertaken to quanititatively validate NavX map to CT registration and determine its suitability for guiding AF ablation. Methods: Sixty-four slice, multidetector CT scans were acquired in 8 dogs and segmented to establish subject-specific LA/PV volumes for importation into the EnSite mapping system. A detailed multi-surface geometry (NavX map) was collected by navigating a mapping catheter in target LA and PVs. Using a novel developmental registration algorithm the map was registered to LA/PV CT volume. Validation was performed using the registered 3D CT model for the guidance of ablation at the location of multiple implanted radio-opaque markers. Results: A total of 39 radio-opaque markers were implanted before CT including 9 LSPV, 8 RSPV, 7 RIPV, 9 LIPV, 3 roof, and 3 LAA locations. Subsequent registration of individual NavX maps to acquired 3D CT volume was successfully completed in all animals. RF ablations were performed at the predicted 3D registered model locations of the radio-opaque markers using 5 mm tip RF ablation catheter. Resulting mean lesion size was 5.3 x 6.5 mm. The lesions enveloped the point of marker attachment to the myocardium in 31/36 (86%) ablations. The average distance from the center of the RF lesion to the targeted site was 2.5 ± 1.5mm. 47 points were successfully navigated under fluorscopic and ICE guidance. All points were accurately portrayed on the 3D registered model. 3 points in the region of the RSPV floor did have minor offsets on the model due to technically difficult segmentation of the RSPV from the right pulmonary artery. Conclusion: These data demonstrate the feasibility of registering NavX derived geometries to CT data sets accommodating sources of cardiac, respiratory, and translational cycle dependent error. Catheter navigation precisely conducted using the registered model demonstrates accuracy within the range required for RF ablation.