Automatic Estimation of Optimal Deployment of Transcatheter Aortic Valve Implantation Using Computed Tomography

A comparison was made between the accuracy of and time saved by using novel automated software for pre-procedural computed tomography (CT) planning before transcatheter aortic valve implantation (TAVI) and manual methods. Preprocedural CT to assess aortic annulus dimensions and predict the optimal C-arm implant angle before TAVI can reduce complications related to incorrect prosthesis sizing and positioning.

A total of 61 consecutive patients underwent TAVI using either the SAPIEN XT or CoreValve prosthesis. Pre-procedural CT scans were analysed using three methods: automatic; semi-automatic; and manual. For each method, annular dimensions were measured and the optimal implantation angle was predicted. After TAVI the actual post-deployment angle orthogonal to the prosthesis was determined using aortic fluoroscopy. The difference between the predicted angle by CT and the measured post-deployment angle was calculated for each method.

For all methods the mean angular difference with the actual post-deployment angle was similar at ~9 ± 7°. There was a significant difference between the SAPIEN XT (6.6 ± 5.8°) and CoreValve (11.5 ± 6.9°, p <0.001) prostheses due to a consistently greater left anterior oblique and caudal angulation for the CoreValve. Although the annular area correlated well among all methods, 'automatic' results were consistently larger than 'manual' results. Interobserver variability was low for all measures. The fully automatic method saved 98 s, and the semiautomatic method 40 s per case.

The use of automatic software enabled a rapid and accurate prediction of implantation angles, though results differed for specific manufacturers. Annular areas were overestimated by the automatic method, and thus required manual adjustments.