Abstract
Transcatheter mitral valve repair (TMVr) is growing in popularity for non-surgical mitral regurgitation (MR) patients, but the manual operation of current TMVr devices increases radiation exposure and limits telesurgery feasibility. A robotically steerable delivery system can alleviate these problems, improving safety and precision while reducing staff fatigue. However, precise manipulation of a surgical robotic system requires system modeling and reliable external feedback. Ultrasound imaging provides visualization and guidance for precise instrument maneuvers within the body. Moreover, it is a readily available, safe, and cost-effective feedback modality, ideal for this procedure. Therefore, in this work, we use a previously derived model for the robotic transcatheter system and perform ultrasound-guided joint space control through real-time (algorithm run time: ~0.011 s) estimation of four joints simultaneously. The joints are estimated using kinematically-derived weight maps, a new technique, and a feature detection algorithm, with an accuracy of 3.19°, 2.76°, 2.41 mm, and 6.83° for the proximal bending, distal bending, prismatic motion, and distal torsion joints, respectively. This approach leverages existing knowledge about the system, demonstrating computational efficiency, intuitive comprehension, and independence from a training dataset, making it a versatile joint estimation technique. Experiments were conducted to compare the proposed method with currently employed joint estimation strategies. Additionally, real-time control was demonstrated using ultrasound feedback in a water bath, while subjecting the robotic transcatheter delivery system to similar tortuosity as encountered during a TMVr procedure.