Date: December 13, 2017
Time: 1:00 PM US/Eastern
Means: In person at Cleveland Clinic (minus one Skype call in)
- Ahmet Erdemir (CC)
- Jason Halloran (CSU)
- Carl Imhauser (HSS)
- Peter Laz (DU)
- Kevin Shelburne (DU)
- Neda Abdollahi (CSU - left early)
- Will Zaylor (CSU - left early)
- Thor Besier (ABI - called in over Skype)
Confirm the understanding of project goals, modeling & simulation stages, and timeline.
- Establish consensus on processes to accomplish individual modeling stages.
- Identify data for use in the project.
Immediate Action Items:
Kevin & Pete (DU)
- Provide short ranked list (at least 2 knees - one male and one female) of possible data sets from and Natural Knee Data.
- Ahmet (CC)
- Provide short ranked list (at least 2 knees - one male and one female) of possible data sets from Open Knee(s).
- Check cartilage grades for Open Knee(s) provided by radiologist.
- Start documenting group-specific protocols for the model development phase.
Based on the agenda items, the group discussed the project workflow, modeling stages and datasets. Ahmet utilized the presentation he gave for the Cleveland Clinic audience as a guide for the meeting (see https://simtk.org/svn/kneehub/doc/CC-2017/presentation_AE.pdf). More detailed notes are provided below.
- The groups discussed what data sets to use for the project:
- One will be from the Natural Knee Data from University of Denver and the other from Open Knee(s). We need one male and one female data set (Open Knee(s) = 4 females, 4 males and Denver has a few knees from male and female donors)
- At a minimum, information from each data set includes imaging and specimen-specific mechanical test results (e.g. joint kinematics-kinetics).
- The discussion centered around using the "highest quality" data from each group.
- Part of each data set should be allocated for validation purposes.
- Because of the use cases (i.e. model reuse), the data sets should have a healthy ACL, and decent cartilage and possibly meniscus.
- Model Development notes.
- The primary deliverables are the "working" knee model and its components. That is, it has not been calibrated; it simply runs.
- The first task will be to document the process to arrive at the "working" model.
- Protocol deviations will be handed with the results (as they become available).
- Model Calibration notes.
- Documentation of group-specific protocols for calibration (algorithms, what data sets are used) will be needed as a first step.
- Model Benchmarking notes.
- We need to decide exactly what parts of each data set will be used for benchmarking.
- There were questions about how benchmarking will be performed, in particular with regards to what will be compared and what type of loading sets (and data) will be used for this purpose. Ahmet discussed earmarking certain data sets for benchmarking (e.g. the ligament cutting data sets from University of Denver and combined loading data sets from Open Knee(s)) and the requirement of "good faith" efforts from the group to utilize the data accordingly.
- Model Reuse notes.
- One item to consider across the groups is how the boundary conditions are applied, especially for the reuse phase. Loading schemes were proposed in the grant, but there could be questions about associated details (location/direction of coordinate frames). We may leave some aspects of reuse, and in particular the boundary conditions, open-ended before more controlled comparisons are made.
- Notes on Operating Procedures (applies to all modeling phases).
- Related activities were listed by Ahmet in the presentation.
- Initial protocols will be delivered and "frozen" in whatever platform is decided on (wiki, data management system, etc.). Throughout execution of the protocols any deviation should be documented. These could be documented on-the-fly, and there should be justification (the "why"). Even what is seemingly an arbitrary deviation should also be documented. Some description is better than none.
- The associated "cost" of each phase should be documented, which includes a number of items to include (development, computational cost, hardware resources, etc.).
- Comparative Analysis notes.
- Third-party evaluators, likely collaborators at the US FDA (arrangements are in progress), will perform the comparative analyses. This will likely be performed at each of the modeling phases (maybe minus the development).
- Each group's job is to submit deliverables in a consistent format (decided by the team consensus and the third-party evaluator) to help with comparative analysis.
- General notes.
- There was some discussion about the possibility, and potential concerns, of cross-fertilization in terms of the group's respective techniques as the project progresses. There's a chance we could converge to the same (or awfully similar) types/approaches near the end.
- Some discussion focused on the timeline of deliverables. It is anticipated that model development should be complete after 1 year. It makes some sense to allocate one year for each of the modeling phases with the first few months in each being used to document the associated processes/deliverables. The timeline of modeling phases can overlap, i.e., when workflow of one is documented and its execution has started, work on documenting the worklow of next can start.
- Ahmet proposed that within the next three months the data sets and deliverables for model development phase will be decided on. Nine months from now, groups will submit their protocols for model development phase, and fifteen months from now the model development phase will be complete (i.e., deliverables from each group will be handed in). This seemed to be an acceptable time line across the group.
- Regarding reporting - at a high level the required detail in terms of reporting each phase was discussed. Reporting standards (ASME, FDA, reporting manuscripts) as well as the Open Knee(s) testing protocols could be used for guidance. Per a question, Ahmet also discussed and presented examples of how the deliverables may be documented. Numerous Open Knee(s) wiki pages were shown, though it is recognized many of these are likely even more detailed than what is needed for this grant (e.g. example results).
- The general "openness" of the project was discussed. It was agreed that external visibility is a product of the grant and it is in the community's best interest for us to be open (and document) about our processes. Related to openness, the possibilities of other groups to join in the process was also discussed. The group agreed that others could participate and are welcome to.