Autonomous System for Vaginal Cuff Closure via Model-Based Planning and Markerless Tracking Techniques

Abstract

Autonomous robotic suturing has the potential to improve surgical outcomes compared to manual operations due to better accuracy and consistency of suture placement. In this letter, we develop an autonomous system with model-based planning and markerless tissue tracking techniques for autonomous vaginal cuff closure. The proposed planning method constructs an optimal suture plan that minimizes tissue collisions with the robotic suturing tool. Moreover, DeepLabCut is utilized for autonomous markerless soft-tissue tracking in robotic vaginal cuff closure. Furthermore, an instruction-based autonomous surgery interface (ASI) is designed to automate the complex suturing workflow. The proposed planning method is evaluated via a robotics simulator and real-world suture-placement tests. End-to-end vaginal cuff closure via Smart Tissue Autonomous Robot (STAR) on synthetic tissues was carried out and compared to the suturing results via conventional laparoscopic and robot-assisted approaches performed by experienced surgeons. The comparison results indicate that by using the proposed planning pipeline, STAR outperforms conventional laparoscopic and robotic-assisted methods with better accuracy and consistency, thus achieving a higher success rate of autonomous suture placement.