Demo implementation of the Mattar 2016 exponential integration model using the forwardModel non-linear fitting platform

The code can be configured using the Matlab package manager Toolbox Toolbox. Alternatively, the following repositories should be downloaded and placed on the Matlab path:

• https://github.com/gkaguirrelab/forwardModel
• https://github.com/freesurfer/freesurfer

There are hard-coded paths to the data and event files in the routine demoMattarAdapt. There are also a couple of parameters there that one could adjust (e.g., spatial smoothing).

Once you have the paths updated, run the demo. The demo script has the flag fitOneVoxel available at the start. Set this to true, and the routine will quickly (in seconds) fit the data for an example voxel. You can use this to confirm that your paths are working properly. Then, set fitOneVoxel to false and run the demo again. The routine will fit the model to all voxels. This takes about 40 minutes on a laptop with 10 CPU cores available for the parpool.

The results of the analysis will be written to the hard-coded save directory. The output includes a set of files in .nii format that provide maps of the relevant parameter values.