Clustering sizing measurements to characterize LAA occluder landing zones

Transcatheter occlusion of the left atrial appendage is an established stroke prevention treatment in patients with non-valvular atrial fibrillation and a contraindication for long-term anticoagulation. However, it entails the risk of major complications and failure to place a left atrial appendage occluder. The large inter-patient variation of the left atrial appendage’s anatomy is a source of risks. Still, no tool for predicting the risks of transcatheter left atrial appendage occlusion due to anatomical variation is available. Consequently, systematically assessing the risks of transcatheter left atrial appendage occlusion based on the measurements available before implantation is not part of current clinical practice. In this retrospective study of 125 patients’ CT-data, 91 datasets fulfilled the inclusion criteria. For the included datasets, the left atrium was segmented, the left atrial appendage was identified, and automated approximation of left atrial appendage occluder sizing measurements was performed. The ability to create distinct groups of landing zone shapes considering the most commonly used morphology classification scheme and the agglomerative hierarchical clustering of the approximated sizing measurements was evaluated. This study demonstrated that the most commonly used morphology classification does not describe landing zones as it is derived from parameters not linked to the left atrial appendage occluder landing zone. Furthermore, this study showed that clustering of sizing parameters for left atrial appendage occluders, without the knowledge of clinical outcomes, yields distinct and clinically relevant clusters. Some of these clusters match known critical left atrial appendage shapes. Therefore, an investigation of landing zone measurements and clinical outcomes will likely identify critical left atrial appendage shapes