T2-Informed-Modeling

This repository provides an outline of the steps, codes and subject parameters to replicate the sensitivity study conducted on T2 informed finite element models by Donabedian et al., PMID #. This study used information from subjects 9948792, 9932809, 9988421, and 9961728 from The Osteoarthritis Initiative (https://nda.nih.gov/oai).

First, obtain the imaging data used in the study.

Download the publicly available hexahedral meshes from Rodriquez-Vila et al.,2017 PMID 29017357.
Download the baseline T2 echo images from subjects 9948792, 9932809, 9988421, and 9961728 from The Osteoarthritis Initiative database.
T2 Maps were calculated in 3D Slicer (5.0.2) and smoothed using an edge preserving gradient anisotropic diffusion filter in SimpleITK (Yaniv, 2018, PMID: 29181613) with five iterations, timestep=0.125, and conductance = 3. The smoothing code is located in the jupyter lab notebook titled T2 Map Smoothing.ipynb. This code also provides the T2 map size, spacing, direction, and origin that are used in the follow up calculations.

Next, we pre-processed all FE simulations in MATLAB (R2024a) using The Geometry and Image-Based Bioengineering add-ON (GIBBON) toolbox. This code is provided by the, T2_informed_FE_modeling.m file. Use this code to create baseline as well as follow up sensitivity simulations with the upper and lower input properties detailed in the Sensitivity Input Parameters.xlsx sheet. The T2_informed_FE_modeling.m code assigns a volumetric weighted T2 value to each element that was calculated from the Volumetric_Weighted_T2.m file. The Volumetric_Weighted_T2.m file requires that the mesh be registered with the T2 Map using the T2_Map_Mesh_Registration.m function which is detailed in the following steps.

Next, calculate the weighted T2 value for each element in the mesh using the The Volumetric_Weighted_T2.m which requires that the mesh be registered to the T2 map using the T2_Map_Mesh_Registration.m function. Ensure that the smoothed T2 map, smoothed T2 map direction, origin and spacing collected in step 3 as well as the rigid affine transformation matrix in the subjects information .xlsx sheet is accessible through the working directory called by T2_Map_Mesh_Registration.m. If done properly, calling the T2_Map_Mesh_Registration.m function should result in a figure similar to the figure titled Registered_picture.png (Note- The registration code converts from the imaging to the cartesian coordinate system. Refer to https://www.slicer.org/wiki/Coordinate_systems for details.)
The Volumetric_Weighted_T2.m code will then to proceed to calculate a weighted T2 value for each element in the mesh based on the overlapping voxels. (Note- This computation can take up to 11 hours to complete.)
Save the weighted T2 array.
Refer back to the T2-informed_FE_modeling.m file to pre-process the rest of the simulation using the baseline input parameters detailed in the Sensitivity Input Parameters.xlsx sheet. This code uses a lot of helpful functions provided by GIBBON to complete. You will need to input spcecific properties used for each subject indicated at various parts in the code. Please use the excel sheet for each subject for this information.
After this code is complete, there will be a respective .feb file. Open this file in FEBio and run this simulation to completion. (Note- These simulations are very sensitive to convergence errors. Please ensure that all the boundary conditions and control parameters are set exactly to how they are provided in each subjects excel sheet.
Once the simulation is complete, all the stress/strain outputs in all elements at each loading step was gathered in FEBio's Data tab.
Refer back to the T2-informed_FE_modeling.m file and change the input parameters based on the modulus shift or altered slope assessment to assess the sensitivity to the input T2-modulus relationship.

Chan-Lab/T2-Informed-Modeling

About

Resources

Stars

Watchers

Forks

Releases

Languages