Automating Vascular Shunt Insertion with the dVRK Surgical Robot

Karthik Dharmarajan, Will Panitch, Muyan Jiang, Kishore Srinivas, Baiyu Shi, Yahav Avigal, Huang Huang, Thomas Low, Danyal Fer, Ken Goldberg

Introduction

In emergency trauma surgery, restoring blood flow to a limb is a critical, time-sensitive task. Surgeons often employ a vascular shunt—a flexible tube—to bridge damaged blood vessels temporarily. Automating this procedure using robotic systems like the da Vinci Research Kit (dVRK) presents significant challenges. These include the inherent deformability of blood vessels, visual occlusion by robotic grippers, mechanical hysteresis in cable-driven systems, and the extremely tight tolerances required for successful insertion.

Methodology

Researchers at UC Berkeley and UCSF developed the Automated Vascular Shunt Insertion (AVSI) pipeline to address these challenges. The system utilizes a bimanual approach combining specialized perception and motion strategies.

The experimental setup using the dVRK to insert a shunt into a vessel phantom.

Perception via LUV

To overcome the difficulty of segmenting vessel rims in complex environments, the team utilized Invisible UV Fluorescent Indicators. By painting vessel rims with UV paint, they could toggle between visible and UV light to generate high-quality "ground truth" labels for training an asymmetric U-Net. This model segments the vessel rim from standard RGB images. The resulting segmentation is then processed using RANSAC (Random Sample Consensus) to fit a 3D circle, determining the vessel's center, orientation, and radius.

The perception pipeline: an RGB input image is processed by a neural network to produce a segmented vessel rim.

Bimanual Coordination and Motion Planning

The AVSI pipeline coordinates two robotic arms to perform the task:

  1. Active Dilation: The first arm grasps and pulls the vessel rim outward to stretch the opening. This step is essential for shunts with diameters close to that of the vessel, as it prevents the rim from collapsing.
  2. Chamfer Tilt and Screw Motion: The second arm performs the insertion. Instead of a direct push, it employs a "chamfer tilt" to catch the edge of the vessel, followed by a "screw motion" (simultaneous rotation and translation). This rotation acts as a mechanical stabilizer, helping the shunt overcome friction and settle inside the vessel without causing the rim to buckle.

Results

The system was evaluated using vessel phantoms at various inclinations (0°, 15°, and 30°). For an 8mm shunt with high clearance, the pipeline achieved a 95% success rate with an average completion time of approximately 13.7 seconds. For a larger 14mm shunt with tight clearance, the success rate was 80% at 0° inclination.

Key findings indicated that active dilation is mandatory for larger shunts; without it, the success rate for the 14mm shunt dropped to 0%. Furthermore, the "screw motion" was identified as a critical factor in maintaining stability during the insertion phase. The system also demonstrated color invariance by training on grayscale data, suggesting potential for generalization to real, blood-stained tissue.

Conclusion and Future Work

The AVSI pipeline demonstrates the feasibility of autonomous vascular shunt insertion with high speed and reliability. However, several hurdles remain before clinical application. Current limitations include the use of simplified latex phantoms and static environments. Future research will focus on improving realism by using tissue-like materials, integrating visual servoing to handle moving targets (such as those held by a human assistant), and automating subsequent surgical steps like ligature knot tying to secure the shunt.

Summary updated

Abstract

Vascular shunt insertion is a fundamental surgical procedure used to temporarily restore blood flow to tissues. It is often performed in the field after major trauma. We formulate a problem of automated vascular shunt insertion and propose a pipeline to perform Automated Vascular Shunt Insertion (AVSI) using a da Vinci Research Kit. The pipeline uses a learned visual model to estimate the locus of the vessel rim, plans a grasp on the rim, and moves to grasp at that point. The first robot gripper then pulls the rim to stretch open the vessel with a dilation motion. The second robot gripper then proceeds to insert a shunt into the vessel phantom (a model of the blood vessel) with a chamfer tilt followed by a screw motion. Results suggest that AVSI achieves a high success rate even with tight tolerances and varying vessel orientations up to 30{\deg}. Supplementary material, dataset, videos, and visualizations can be found at https://sites.google.com/berkeley.edu/autolab-avsi.
BibTeX
@inproceedings{dharmarajan2023automating,
  author = {Karthik Dharmarajan and Will Panitch and Muyan Jiang and Kishore Srinivas and Baiyu Shi and Yahav Avigal and Huang Huang and Thomas Low and Danyal Fer and Ken Goldberg},
  title = {{Automating Vascular Shunt Insertion with the dVRK Surgical Robot}},
  booktitle = {IEEE International Conference on Robotics and Automation},
  year = {2023},
  url = {https://openreview.net/forum?id=FJz0ySxQ9Y},
}
Figure