Project Page: Retinal Vein Cannulation

Retinal vein occlusion is one of the most common causes of vision loss, occurring when a blood clot or other obstruction occludes a retinal vein. A potential remedy for retinal vein occlusion is Retinal Vein Cannulation (RVC), a surgical procedure that involves infusing the vein with an antifibrinolytic to restore blood flow through the vascular lumen. This work presents an image-guided robotic system capable of performing fully automated retinal vein cannulation on silicone phantoms. The system is integrated with an Optical Coherence Tomography (OCT) probe and camera to provide visual feedback to guide the robotic system. The developed system was evaluated through a set of experimental trials and shown to be capable of performing automated retinal vein cannulation with no surgical complications. The developed technology is expected to lead to improvements in vitreoretinal surgical procedures such as subretinal injection or other tech- niques which are difficult or infeasible to perform with traditional methods.

  • This project used the existing IRISS and integrated Optical Coherence Tomography (OCT) imaging system. In this work, this system was used to demonstrate automated, image-guided retinal vein cannulation on silicone vein phantoms.

  • This is a system-level diagram showing data and signal flow. The OCT and camera data are used as visual feedback in the automatic controller (host PC) to control the micropipette.

  • I designed and fabricated a custom piezo-actuated tool holder to mount and control the glass micropipette. Dimensions are indicated: Note the small size of the micropipette, necessary for cannulating the retinal vein phantoms.

  • Shown is an illustration of the five-step process to create the retinal vein phantoms. Aside from an overnight cure time, the process was fast (approximately 10 minutes) and produced sufficiently realistic vein phantoms.

  • (a) In the first step, the operator selects a site in the camera image for the system to cannulate. All subsequent steps are fully automated and require zero operator input. (b–d) An OCT B-scan is acquired through the vein and automated image processing is employed to detect the vascular lumen.

  • For guidance, the system automatically generates a virtual model of the eye and plans trajectories to track. The trajectory consists of two straight-line approaches: a portion to align the mechanical remote center of motor (RCM) of the robotic system to the virtual scleral incision point and a second to bring the tip of the micropipette near the desired cannulation point.

  • Here is a schematic that shows the three main steps of the procedure. Solid black dots represent start points; white dots represent end points. Variable definitions are provided in the associated journal publication (below).

  • This illustration shows the mathematical definitions during the vein targeting step. Note: Relative dimensions are not to scale.

  • The system automatically confirmed vein cannulation by detecting a color change (green to red) in the glass micropipette after vein puncture. This color change was guaranteed upon successful cannulation due to the controlled pressure differential between the blood analog inside the vein phantom and the dyed water inside the micropipette. Important steps of the image-processing algorithm included (a) selecting an ROI, (b) blurring and masking, and (c) averaging hue values to determine color.

  • (a) Side view and (b) top view of OCT data demonstrating OCT-based visual servoing of the micropipette during the vein-targeting step. (c) The calculated error shows a fast (two iteration) decrease to less than the error threshold (20 µm).

  • (a & b) B-scan data evaluated to detect the optical shadow cast by the micropipette, (c) detected micropipette centerline in the top view, and (d) detected steel edge, micropipette tip, and micropipette centerline in the side view

  • Two examples of the OCT feedback provided to the operator at a rate of 5 Hz during automated retinal vein cannulation of (a) a ⌀120 µm vein and (b) a ⌀160 µm vein. The tip of the micropipette (yellow "plus" symbol) was automatically overlaid atop the data and followed the tip as it cannulated the vein. In (b), the example shown was acquired within 1 s of infusion and shows the dyed water (yellow arrow) being infused through the vein phantom.

Publications

  • Automated Retinal Vein Cannulation on Silicone Phantoms using Optical Coherence Tomography–Guided Robotic Manipulations, M.J. Gerber, J.P. Hubschman, and T.C. Tsao, ASME/IEEE: Transactions on Mechatronics (TMECH), Dec. 2020 – DOI | PDF
  • Optical Coherence Tomography–Guided Robotic System for Automated Retinal Microsurgery, M.J. Gerber, PhD Dissertation, University of California, Los Angeles (2019) – Permalink | PDF

Resources

© 2019–2021 Matthew J. Gerber gerber211@ucla.edu A-114 Jules Stein Building, Los Angeles, CA