• RecurringMeetings
• 2019-07-09

Recurring Meeting of Cleveland Clinic Core Team

Date: July 9 2019

Time: 2:00 PM EST

Means: In person meeting

Attendees:

1. Ahmet Erdemir
2. Ellen Klonowski
3. Sean Doherty
4. Benjamin Landis (phone)

Agenda:

1. Model assembly manuscript.
2. Lumped in vitro models.
3. Morphing for lumped in vivo models.
4. Layered in vitro models.
5. Tissue testing.
6. Recap of action items.
7. Other.

Ongoing Action Items:

• Rici
  • Evaluate coordinate systems for in vitro surgical tools manuscript.
  • Prepare in vivo indentation manuscript for resubmission.
• Ben
  • Upload latest version of model assembly manuscript for Ahmet to review.
  • Meet with Ahmet to freeze and write up morphing strategy.
• Sean
  • Rerun and calibrate lumped models for publication.
  • Meet with Ahmet to learn and write up new layered modeling strategy.
  • Meet with Ellen to discuss data resampling.
• Ellen
  • Locate cartilage data to test scripts on zero stress-strain point identification.
  • Meet with Sean to discuss data resampling.
• Ahmet
  • Meet with Ben to freeze and write up morphing strategy.
  • Meet with Sean to demonstrate new layered modeling strategy.

Notes:

1. Model assembly manuscript.
   • Ben looked at the manuscript a little bit; he will put it up what he has. Ahmet will look at it and get it out of the way this summer.
2. Lumped in vitro models.
   • Sean presented a poster at the SB3C and talked to people about these models. He has a few things to tweak, i.e., what we do with the probe. Sean wrote the scripts to process data and rerun simulations. Previously, Sean assumed that the probe didn't rotate and the initial orientation was set constant during its movement. We now use the average orientation of the probe during experimentation. Also, rather than going from start point to end point as a straight linear displacement; we now fit a line to all translations and use that as our probe displacement direction. Any errors in Optotrak can be accommodated using deviation from this line as a metric to include the experimental point.
3. Morphing for lumped in vivo models.
   • Ahmet and Ben previously talked about handling morphing. The discrepancy in scaling based on circumference and morphing based on tissue thicknesses was an issue. Ben was not sure if Ahmet thought about any further than before. Ahmet recommended to morph for the thicknesses at four locations: anterior, posterior, medial, lateral. For the cadaver model (template), we have ultrasound thickness measurements. For the target patient model, we have ultrasound thickness measurements. We calculate the ratio of thickness change from these and apply them to locate morphed point on skin surface where they were measured. We need iterate the workflow in writing. The overall goal is to simplify the process and freeze it. Ahmet and Ben will talk in their weekly meeting to harden how we do it.
4. Layered in vitro models.
   • We have three different types of models. First type of models having an additional surface layer at the interface of tissue, which is smoothed, cut, and have perfect conformity. In this type of modeling, problems in geometric representation rise, i.e., with regions with two bones and when some layers are thin. The second type of modeling strategy relies on adaptive meshing and assignment of material property based on its voxel. The approach works but the intermediate surfaces are not smooth and mesh size requires a lot of memory to run. Currently there are four upper extremity models with the first strategy and an upper leg model with the second strategy. Third type of modeling relies on joint smoothing of tissue layers during model generation in Slicer. This ensures conformity of intermediate surfaces. Without decimation, a uniform but dense surface mesh can be obtained. These surfaces can then be meshed in SALOME to get volumetric meshes. Ahmet already performed some meshing with this workflow. The mesh sizes will likely be comparable to those of adaptive meshing. These meshes were not tested in FEBio yet. Model simplification based on simulation of a region of interest (implemented through Ariel's model reduction script) will likely decrease memory requirements and computational cost. Ahmet and Sean will discuss to freeze this third modeling strategy, write up the process and test.
5. Tissue testing.
   • Ellen added RMS calculation for stress and strain curve. This can help compare ramp 1 and ramp 2 (before and after pre-conditioning) and also provide a measure of repeatability of the tests. Ahmet asked if Ellen was doing any resampling as strain range and strain values of the data may not exactly align. Ellen and Sean will talk about resampling strategies as Sean has to do it for lumped model calibration. Ellen can manually adjust the inflection point. With rubber samples she was having a hard time to identify zero stress-strain as inflection point is difficult to register due to compressive stiffness. She was asking if there were any data on skin to look at. Ahmet recommended to check Snehal's data on cartilage. Ellen will check in-house data management system to find these. Ahmet also recommended that she may want to do compression testing as data from such tests will have definitive inflection point when force accumulates with contact.
6. Recap of action items.
   • The participants went over the action items.
7. Other.
   • Ahmet mentioned that next modeling steps will be layered morphed and lumped and layered models in SOFA.