A Novel 3D Segmentation Technique for Patient-Specific Aortic Valve Modeling

Abstract

Accurate three-dimensional modeling of the aortic valve is essential for understanding patient-specific anatomy, guiding surgical and interventional planning, and improving outcomes in the management of structural heart disease. The complex, dynamic geometry of the aortic valve—particularly the delicate and highly variable leaflet morphology—presents significant challenges for existing segmentation methods, especially in the presence of calcifications or congenital anomalies. We present a novel, image-based manual segmentation technique that employs spline interpolation on computed tomography (CT) images to identify and delineate the aortic valve leaflets with high anatomical fidelity. The method involves the guided generation of cubic splines on orthogonal image planes: for each leaflet, splines representing the body are first constructed, followed by splines outlining the fixed and free edges. These splines are then integrated to reconstruct the leaflet surface. A three-dimensional model is subsequently generated by assigning the measured leaflet thickness to the reconstructed surface. This approach provides a reproducible and anatomically accurate model of the aortic valve, particularly valuable in cases where automated segmentation methods fail or require manual correction. Preliminary application to clinical datasets of 25 patients and one normal aortic valve patient demonstrates the method’s potential for precise valve reconstruction and its applicability in preoperative simulation, valve-sparing surgical procedures, and transcatheter aortic valve interventions (TAVI). Additionally, the technique holds promise for supporting accurate fluid–structure interaction (FSI) simulations. By bridging manual expertise and digital modeling, this method supports improved patient-specific treatment strategies. Further development toward automation is anticipated.