Manual, semiautomated, and fully automated measurement of the aortic annulus for planning of transcatheter aortic valve replacement (TAVR/TAVI): Analysis of interchangeability

• Preprocedural CT angiography allows dedicated 3-dimensional reconstructions of the aortic annular plane. However, manual 3-dimensional reconstructions are complex, with the potential for interobserver and intraobserver variability.
• We compared fully automated, semiautomated, and manual aortic annulus measurements in patients evaluated for transcatheter aortic valve replacement and evaluated the implications of the analysis method on recommended valve size.
• Our results demonstrate that semiautomated aortic annulus analysis, with minor correction by the user, provides reliable results in the context of transcatheter aortic valve replacement annulus evaluation.
• Future studies will need to address potential impact on device selection and procedural outcome.

BackgroundPreprocedural 3-dimensional CT imaging of the aortic annular plane plays a critical role for transcatheter aortic valve replacement (TAVR) planning; however, manual reconstructions are complex. Automated analysis software may improve reproducibility and agreement between readers but is incompletely validated.MethodsIn 110 TAVR patients (mean age, 81 years; 37% female) undergoing preprocedural multidetector CT, automated reconstruction of the aortic annular plane and planimetry of the annulus was performed with a prototype of now commercially available software (syngo.CT Cardiac Function—Valve Pilot; Siemens Healthcare, Erlangen, Germany). Fully automated, semiautomated, and manual annulus measurements were compared. Intrareader and inter-reader agreement, intermodality agreement, and interchangeability were analyzed. Finally, the impact of these measurements on recommended valve size was evaluated.ResultsSemiautomated analysis required major correction in 5 patients (4.5%). In the remaining 95.5%, only minor correction was performed. Mean manual annulus area was significantly smaller than fully automated results (P < .001 for both readers) but similar to semiautomated measurements (5.0 vs 5.4 vs 4.9 cm2, respectively). The frequency of concordant recommendations for valve size increased if manual analysis was replaced with the semiautomated method (60% agreement was improved to 82.4%; 95% confidence interval for the difference [69.1%–83.4%]).ConclusionsSemiautomated aortic annulus analysis, with minor correction by the user, provides reliable results in the context of TAVR annulus evaluation.