Automatic Estimation of Optimal Deployment Angles and Dimensions for Transcatheter Aortic Valve Implantation: Comparison between Edwards Sapien XT and Corevalve Prostheses

Introduction: Preprocedural computed tomography for assessment of aortic annulus dimensions and prediction of optimal C-arm implant angle before transcatheter aortic valve implantation (TAVI) can reduce complications, related to incorrect prosthesis sizing and positioning, and minimizes the use of intravenous contrast during TAVI. We aimed to evaluate the accuracy and measure the time saved by use of novel automated software programs for preprocedural planning compared to manual methods.

Methods: 61 consecutive patients underwent TAVI using either Sapiens XT (n=31) or Corevalve (n=30). Preprocedural CT scans were analyzed by each of 3 methods: automatic, semiautomatic and manual. Annular dimensions, and the line perpendicular to the annulus (‘line of perpendicularity’) were used to derive the predicted implantation angle for each method. After the TAVI procedure, the actual post-deployment angle orthogonal to the prosthesis was determined using fluoroscopy and the difference from the predicted angle was calculated.

Results: For the automatic method, the mean angular difference was 9.1 ±6.8°, very similar to results of the manual and semiautomatic methods. There was a significant difference between Sapiens XT (6.6 ±5.8°) and Corevalve (11.5 ±6.9°, p<0.001), due to a consistently higher LAO angulation for Corevalve. The annular area correlated well between all methods, however automatic results were consistently larger than manual results by a factor of 5% in diameter and 10% in area. Interobserver variability was low for all measures. The fully automatic method saved 98 seconds and the semiautomatic method 40 seconds per case.

Conclusions: Use of automatic software enables speedy and accurate prediction of implantation angles. Corevalve yields a higher LAO configuration post implantation, probably related to distortion caused by traction on the distal part of the stent in the ascending aorta. Annular areas are overestimated by the fully automatic method, but editing the automatic contours still results in significant time-saving.