This is the development site of Reproducibility in simulation-based prediction of natural knee mechanics. The development efforts for the project is through a collaboration between Ahmet Erdemir (Cleveland Clinic), Jason Halloran (Cleveland State University), Peter Laz & Kevin Shelburne (University of Denver), Carl Imhauser (Hospital for Special Surgery) and Thor Besier (Auckland Bioengineering Institute). The project is currently funded by the National Institute of Biomedical Imaging and Bioengineering, National Institutes of Health (Grant No. R01EB024573).
The specific aims are
- To quantify the influence of variations in modeling and simulation workflows on the reproducibility of joint level predictions in computational knee biomechanics.
- To quantify the influence of variations in modeling and simulation workflows on the reproducibility of tissue level predictions in computational knee biomechanics.
- Project Overview
Project website - https://simtk.org/projects/kneehub
Project wiki - https://simtk.org/plugins/moinmoin/kneehub/FrontPage
Source code repository - https://simtk.org/svn/kneehub/
- Data management site
Erdemir A, Besier TF, Imhauser CW, Laz P, Morrison T, Shelburne K, Halloran JP. A collaborative pathway to establish credible practice of modeling and simulation in knee biomechanics in conformance with community recommendations. 2018 IMAG Futures Meeting – Moving Forward with the MSM Consortium, March 21-22, 2018, Bethesda, MD. Abstract Poster
Reviews & Perspectives
Erdemir A, Hunter PJ, Holzapfel GA, Loew LM, Middleton J, Jacobs CR, Nithiarasu P, Löhner R, Wei G, Winkelstein BA, Barocas VH, Guilak F, Ku JP, Hicks JL, Delp SL, Sacks M, Weiss JA, Ateshian GA, Maas SA, McCulloch AD, Peng GCY. Perspectives on Sharing Models and Related Resources in Computational Biomechanics Research. J Biomech Eng. 2018 Feb 1;140(2). doi: 10.1115/1.4038768. PubMed
- Cleveland Clinic 2017
Presentation by Ahmet Erdemir - https://simtk.org/svn/kneehub/doc/CC-2017/presentation_AE.pdf
https://simtk.org/svn/kneehub/doc/project_summary.pptx - a two-slide summary of the project
https://simtk.org/svn/kneehub/doc/grant_resubmission.pdf - grant narrative (resubmission)
https://simtk.org/svn/kneehub/doc/CC-2017/presentation_AE.pdf - a detailed presentation on the workflow (from Cleveland Clinic 2017 gathering)
Each modeling & simulation phase will have data processing and comparative analysis stages. Items above on data and comparative analysis is for the whole project.
Processes should be defined for deposition & dissemination of data, specifications, protocol deviations & deliverables.
- Grant narrative provides guidance for the overall project workflow.
Context of Use of Models
For target clinical and research use cases (see model reuse):
- Prediction of tibiofemoral and patellofemoral joint kinematics-kinetics
- Prediction of tissue load sharing, individual tissue loads, and stress-strain
- Open Knee(s) - from Cleveland Clinic
- oks003 - 25 years old, female donor; left knee
- Natural Knee Data - from University of Denver
- DU02 - 44 years old, male donor; right knee
For more details on data resources and candidate specimens, refer to DataResources.
Modeling & Simulation Phases
Overall Strategy to Document and Execute M&S Phases
Each modeling & simulation phase of the project will be staged:
- Define primary and secondary deliverables (by group consensus)
- Define earmarked data (by group consensus)
- Define timeline (by group consensus)
Curate earmarked data (by Cleveland Clinic & University of Denver teams)
- Prepare earmarked data (to bring it in a usable form for exchange) (by group consensus)
- Prepare specifications (by individual teams); this may need to reflect availability of data in two different data groups
- Submit specifications (by individual teams)
- Review specifications (by group consensus for adequacy of detail)
- Execute specifications (by individual teams)
- Document protocol deviations (by individual teams)
- Submit protocol deviations (by individual teams)
- Review protocol deviations (by group consensus for justification and for adequacy of detail)
- Submit deliverables (by individual teams)
- Review deliverables (by group concensus for completeness)
- Prepare package for comparative analysis (by Cleveland Clinic team)
- Comparative analysis (by third-party)
Each individual step will require expected detail and guidance to accomplish that expected detail. Individual phases will likely have time overlap, i.e., while specifications are executed for one phase, they can be developed for the next phase.
Model Development Phase
Goal. (each modeling team) To develop two initial working models (one from Open Knee(s); one from Natural Knee Data) - starting with earmarked data and providing all deliverables.
Earmarked Data. Specimen-specific medial imaging datasets (MRIs, CTs), other specimen-specific anatomical information (probing), literature.
Deliverables. (for each model) working model simulating a sample scenario, intermediate and final virtual representations of model components, sample simulation results; documentation of modeling and simulation processes - specifications (prior to execution), protocol deviations (changes in specifications during and posterior to execution)
Details. See ModelDevelopment.
Model Calibration Phase
Goal. To develop two calibrated working models (one from Open Knee(s); one from Natural Knee Data) - starting with earmarked data and providing all deliverables.
Earmarked Data. Specimen-specific joint mechanics datasets (joint kinematics-kinetics during laxity testing), other specimen-specific anatomical information (probing data during mechanical testing), literature.
Deliverables. (for each model) calibrated working model simulating the same sample scenario of the previous phase, simulation results of the sample scenario; for each calibration stage - calibrated parameters (before and after), representation of loading cases selected from earmarked data for calibration, simulation results of loading cases used for calibration (before and after), calibration fit error (before and after), intermediate and final virtual representations of model components that are changed during calibration; representation of all loading cases of earmarked data, simulation results of all loading cases of earmarked data with calibrated model; documentation of modeling and simulation processes - specifications (prior to execution), protocol deviations (changes in specifications during and posterior to execution).
Details. See ModelCalibration.
Model Benchmarking Phase
Goal. Quality of tuned model (to reference and relative, e.g. validity domain)
Details. See ModelBenchmarking.
Model Reuse Phase
Goal. Predictive potential of tuned model in extrapolation (to each relative, e.g. applicability)
Details. See ModelReuse.