Abstract A Level 2 autonomous robotic system for coronary interventions could replace clinicians in routine tasks, leading to more efficient, consistent and safer procedures. However, due to the unknown 3D instrument position, the unawareness of the current surgical status, and the inability to understand intraoperative coronary arterial angiography (CAG) images, robots struggle to achieve surgical tasks independently and safely. Here, we demonstrate a Level 2 autonomous surgical robotic system that conducts specific tasks of a standardized coronary intervention. The proposed constrained spatial position probability prediction (CS3P) algorithm can implement millimeter-level 3D instrument localization, and the designed dynamic perception and adaptive control (DPAC) strategy could accurately predict delivery status and adopt human-like operational skills to help smooth delivery. Fast and accurate analysis of CAG images based on deep learning is provided for diagnosis, monitoring and assessment. Exhaustive experiments on both living animals and vascular models prove that our robot can autonomously perform specific tasks with high precision and stability, promising to achieve standardized, safe and efficient coronary interventions.
Hou et al. (Mon,) studied this question.