Biography

Bo Lu received the B.Eng. degree from Dalian University of Technology, China, in 2013. He then obtained his M.S. Degree (first honor) and Ph.D. Degree from the Department of Mechanical Engineering, The Hong Kong Polytechnic University, Hong Kong SAR, in 2015 and 2019, respectively. He is now a Post-Doctoral Research Fellow in the T-stone Robotics Institute, The Department of Mechanical and Automation Engineering, The Chinese University of Hong Kong, N. T., Hong Kong SAR. His research interests include medical robotics, computer vision, vision-based manipulation, automation and control, task planning, surgical intelligence, surgical procedure analysis and understanding.

Abstract

Surgical suturing, one of the most fundament yet a tedious and time-consuming procedure, plays an important role in surgery that a high-quality suturing can significantly benefit the postoperative recovery for patient. Meanwhile, it brings the problems that surgeons need to highly concentrate on surgical manipulation, and longtime operation and consistent pressure can easily result in fatigue when conducting the wound suturing.

To further assist surgeons and release them from tedious procedures, vision-based approaches towards automated surgical wound suturing were consequently proposed with the aid of surgical robotic system. In this report, we delicately investigate the procedures towards automated suture thread cutting, grasping, and the looping manipulation among the overall surgical suturing manipulation. To resolve the challenges concerning suture thread manipulation, we summarize our studies of suture thread segmentation and the corresponding 3D shape reconstruction based on both the classical methods (e.g. contour-based segmentation, multiple image operators-based filtering) and learning-based approaches with our own surgical data. Accordingly, we also give insights into online surgical environment reconstruction by using Thin Plate Spline (TPS) model, which can remarkably improve the efficiency of automated suture thread grasping task. In addition, we also discuss a dynamic suture thread looping method through efficient trajectory planning, real-time visual perception, and a Linear Quadratic (LQ) controller using an in-house robotic system, and potential challenges, such as suture slippage and collisions between instruments, can be efficiently eliminated. With the well-established workflow of vision-based surgical suture thread manipulation, we will address its integration to the da Vinci Research Kit (dVRK), as well as looking to the vision of realizing the overall automated robotic surgical suturing.