Table of Contents

Building a Workflow for Cooperatively Controlled Robotic Mandibular Surgery

Last updated: 05/11/2022 10:25am

Summary

Mandibular osteotomy is a common procedure used to correct an overextended jaw or receding chin (see Figure 1) [1]. This is done by making a sagittal cut through either side of the mandible as shown in Figure 2 below. During this procedure, the physician must be careful not to damage the alveolar nerve that runs through the mandible [3]. Damage to this nerve can lead to numbness of the chin, lower lip, and lower teeth. According to Dr. Yang, 100% of patients experience a temporary neurosensory deficit following the procedure with 10% of those cases being permanent.

Mandibular osteotomy is a common procedure used to correct an overextended jaw or receding chin [1]. This is done by making a sagittal cut through either side of the mandible as shown in Figure 1 below. During this procedure, the physician must be careful not to damage the alveolar nerve that runs through the mandible [3]. Damage to this nerve can lead to numbness of the chin, lower lip, and lower teeth. According to Dr. Yang, 100% of patients experience a temporary neurosensory deficit following the procedure with 10% of those cases being permanent.

Background, Specific Aims, and Significance

Background

For the BSSO procedure, current standard practice is for the surgeon to execute the cut by hand. A flowchart identifying a typically workflow is shown in Figure 3 below.

To begin, a CT scan is taken of the patient’s head and a surface scan is taken of the patient’s teeth. This is done because many of the patients typically have braces, which will create artifacts in the CT image. The vendor will take the scans and merge them together while producing the segmentation. Using the segmented model, the vendor and the surgeon will create a plan for the BSSO. The vendor will produce splits, brackets, and plates to be used in the procedure. Splints are used to line up the teeth in the desired final position after the cut has been made and before screwing the mandible into place. Brackets and plates are screwed onto the mandible to fix the cut bone into its final position. Once the plan, brackets, and splints are made the physician will execute the procedure by hand by using a drill to cut through the bone.

Specific Aims and Significance

Deliverables

Technical Approach

Mandible Phantoms

The first portion of this project focused on the development of a mandibular phantom. This will be needed for testing the system and verifying its performance. Developing such a phantom has additional applications in physician training models and creating other bone phantoms such as a temporal bone for a mastoidectomy.

Fabrication of these phantoms was done using a form of 3D printing called Digital Light Processing (DLP). This process works by curing UV resin one layer at a time with a UV LCD panel. DLP is able to create accurate phantoms from CT segmentations which will preserve anatomical structures better then a casting in dental stone would. Photopolymer resin, unlike thermoplastics used in traditional Fused Deposition Modeling (FDM) 3D printing, does not melt when machined. Additionally, elastic and brittle resins can be mixed to customize the physical properties of the material. Using this as a basis, a mandible phantom can be created that will accurately reproduce patient anatomy and represent the machinability of bone (Figure 5).

One issue in accurately representing machining bone with these prints is that cancellous bone is not created in the model. This is because the CT segmentation is typically intensity based which is correlated with bone density. Porous structures such as cancellous bone will show up as a lower intensity in the CT images, but even if it was included in the segmentation, it would be printed the same as the cortical bone which would result in one solid phantom which does not accurately represent bone. Another problem that was encountered was the cortical bone in the segmentation showed up thicker than what the physicians expected as shown in Figure 6.

Two adjustments were made to the model to address these issues. First, the segmented mandible model was shelled out to create an artificial 1mm thick cortical bone. This thickness was selected from a paper that identified the thickest the cortical bone in the mandible is likely to be is 1mm [6]. Second, to represent cancellous bone a lattice structure was generated to fill the internal geometry of the mandible. Both of these features can be seen implemented in the image below.

The lead mentor for this project, Dr. Yang, drilled into these phantoms and verified that their machinability was representative of real bone. A detailed procedure on how these phantoms were made can be found in the documentation section.

Clinical Workflow

Another aspect of this project is to design a method for implementing the robotic system into the clinical workflow. The current workflow that was shown above in Figure 3 lends itself well to robotic implementation. See the figure below for an outline of the proposed procedure.

Parts of the workflow that will be changed from the current procedure are highlighted in the figure. Because segmented models and digital surgical plans are already created for the procedure, very little of what the surgeon does will need to be changed. The surgeon’s plan can be used to automatically generate virtual guidance that will be enforced by the robot during the procedure. When the vendor creates splits that fit onto the teeth for alignment, a mount for the tracking markers can be created that will fit over the teeth as shown in Figure 9. This will be used so the robot can track the location of the mandible during the procedure. While drilling the physician will operate from one side of the patient with an assistant standing near the patient’s head. The robot will reside across from the physician and hold onto the tool with the surgeon throughout the procedure which allows it to enforce the pre-defined plan. An example of this layout is shown in the figure below.

The aspect of this workflow that is likely to change as the project evolves in how tracking is done during the procedure. As mentioned before, the YOMI robot uses a tracking arm instead of optical markers. This eliminates the risk of the physician accidentally obscuring the markers while operating in the oral window. Given the small operating area, blocking the markers will likely be a problem. Extending the length of the mount’s neck could improve marker visibility but will reduce accuracy the further the markers are from the mandible. For now, the project will proceed by using the optical tracking markers to establish the feasibility of the workflow.

Bench Testing Workflow

Before building the clinical system, this workflow will need to be tested on the bench. A similar procedure was designed for testing the proposed workflow which is shown in Figure 10 below.

To begin, a phantom mandible is screwed into the mounting plate as shown in Figure 11. There is a cutout in the back of the mandible, so it fits securely on the mount. From here the setup can be scanned using Cone Beam CT (CBCT) to register the relationship between the phantom and the optical tracking markers. The markers are laid out in such a way as to represent where the tracking markers would be in the clinical workflow if using the teeth mounted optical markers. Additionally, markers are spaced out as far as possible to reduce tracking error. Tracking markers will also be located on the drill as shown in Figure 12. The drill will be used to perform a pivot calibration using the divots in the corners of the mounting plate to register the relationship between the tip of the tool and the tool’s markers.

An adaptor was designed to mount the drill to the Galen robot’s end-effector which can be seen in the figure. 3D Slicer can then be used to generate a planned planar cut for the procedure similar to what was done in the Galen Mastoidectomy paper [5]. The Galen can then be used to execute the plan on the mandible phantom. After the operation, the phantom will be scanned again to compare the cut to the plan and measure its deviation. This will be used to compare the effectiveness of the cooperative robotic method to a free-handed operation.

Dependencies

Milestones and Status

  1. Milestone name: Mandible phantoms
  2. Milestone name: Galen configured for mandibular osteotomy
    • Planned Date: 04/08
    • Expected Date: 05/03
    • Status: In progress
  3. Milestone name: Mandible registration with Galen
    • Planned Date: 04/22
    • Expected Date: 06/07
    • Status: In progress
  4. Milestone name: Virtual Fixtures Algorithm
    • Planned Date: 06/03
    • Expected Date: 08/09
    • Status: Not started
  5. Milestone name: Workflow Documentation
    • Planned Date: 07/10
    • Expected Date: 08/109
    • Status: Not started

Reports and Presentations

Project Bibliography

Other Resources and Project Files

Meeting Minutes

Documentation