Toward automated tissue retraction in robot-assisted surgery

Robotic surgical assistants are enhancing physician performance, enabling physicians to perform more delicate and precise minimally invasive surgery. However, these devices are currently tele-operated and lack autonomy. In this paper, we present initial steps toward automating a commonly performed surgical task, tissue retraction, which involves grasping and lifting a thin layer of tissue to expose an underlying area. Given a model of tissues in the vicinity, our method computes a motion plan for a 6-DOF gripper that grasps a tissue flap at an optimal location and retracts it such that an underlying target is fully visible. The planner considers three optimization objectives relevant to medical applications: minimizing the maximum deformation energy, minimizing maximum stress, and minimizing the control effort in lifting the tissue flap. The planner can be used to locally improve physician specified retraction trajectories based on the optimization criteria or to compute a de novo plan. We use a physically-based simulation to compute equilibrium configurations of the tissue flap subject to manipulation constraints. These configurations are used with a sampling-based planner to explore the space of deformations and compute an optimal plan subject to discretization and modeling error. Our experimental results illustrate the ability of the method to compute retractions for heterogeneous tissues while avoiding obstacles and minimizing tissue damage.