Autonomous Patient-Specific Geometry-Aware Planning and Control Framework for Robotic Transcatheter Tricuspid Valve...

Tricuspid regurgitation (TR) is the most prevalent yet historically neglected valvular disease. Transcatheter tricuspid valve interventions (TTVI), encompassing both repair and replacement, have emerged as less invasive alternatives, but their clinical scalability is hindered by the challenge of navigating a large-diameter catheter within the anatomically complex right atrium. This work aims to establish an autonomous and patient-specific planning-control framework that bridges preoperative imaging with intraoperative robotic navigation.

Cardiac chambers were reconstructed from preoperative computed tomography (CT) segmentation. A singular value decomposition-based valve analysis was applied to automatically extract the tricuspid annulus and estimate its centerline. On this anatomical foundation, a geometry-aware path optimization method was developed, integrating signed distance fields (SDF) with Riemannian metrics to generate safe, smooth, and anatomically consistent delivery paths. During intraoperative navigation, a numerical Jacobian-based inverse kinematics controller guided the catheter tip autonomously along the preplanned path.

The framework was validated in a preclinical setting using fifteen real patient CT datasets with severe TR and three cardiac phantoms. The proposed method consistently produced anatomically valid paths and achieved accurate and robust catheter tip tracking, with millimeter-scale tracking errors across repeated phantom experiments.

The integration of singular value decomposition-based anatomical analysis, geometry-aware path planning, and Jacobian-based control enables autonomous catheter navigation directly derived from patient-specific CT data.

A framework was developed for TTVI to establish an end-to-end pipeline ranging from patient-specific CT analysis to autonomous catheter navigation, thereby reducing the burden on the operator and charting a course that provides a foundation for future translational development in robot-assisted structural cardiac interventions.