Comparative Study of Automated Real-Time Left Atrial Appendage Sizing Using Patient-Specific 3D Heart Models Versus Transesophageal Echocardiography and Multidetector Computed Tomography in Patients with Nonvalvular Atrial Fibrillation: Implications for Device Selection

Background: An accurate device sizing for percutaneous left atrial appendage closure (LAAC) can be challenging. Intraprocedural automated LAA evaluation by transoesophageal echocardiography (TEE) based on an artificial intelligence-assisted 3D model may facilitate sizing and prediction of C-arm angulation for device implantation in a one-stop-shop procedure. This study aimed to evaluate the feasibility and accuracy of automated echocardiographic LAA sizing based on a patient-specific 3D heart model prototype in real-time. Methods: A prospective monocentric study was conducted in 66 patients who underwent LAAC with the Amulet device. All major LAA morphologies were represented. Preprocedural multidetector computed tomography (MSCT) measurements and intraprocedural TEE and angiography measurements of the ostium, landing zone, and C-arm angulation were compared with the 3D heart model measurements. Results: The procedure achieved a 100% success rate. The measurements for the maximum diameter of the ostium in the 3D heart model were not significantly different from those obtained via angiography, TEE, and MSCT. Specifically, the maximum diameter of the landing zone did not differ significantly compared to TEE and angiographic measurements (20.90 ± 3.42 mm vs. 20.96 ± 4.81 mm, p = 0.563; compared to 21.20 ± 3.90 mm, p = 0.291). However, these measurements were significantly smaller than the average MSCT measurements (18.30 ± 2.68 mm vs. 21.03 ± 4.34 mm, p < 0.001). Additionally, the predicted implantation angles showed no significant differences between the 3D heart model and MSCT. Conclusions: The intraprocedural application of this prototype is both safe and feasible. The measurements obtained from the 3D heart model were consistent with those from TEE and angiography, although discrepancies were noted when compared to MSCT measurements. Notably, the predicted implantation angles demonstrated strong agreement with MSCT, further supporting the prototype’s efficacy in clinical settings.