Code for running simulations and data analysis from:
Cole MW, Ito T, Schultz D, Mill R, Chen R, Cocuzza C (2019). "Task activations produce spurious but systematic inflation of task functional connectivity estimates". NeuroImage. doi:10.1016/j.neuroimage.2018.12.054 https://doi.org/10.1016/j.neuroimage.2018.12.054
A preprint version of the article is freely available here: https://www.biorxiv.org/content/10.1101/292045v3
Minimal model: A simple model with easy-to-read Python code to demonstrate the basic activation-based task-state FC inflation effect and an effective way to correct the issue.
Neural mass model: Jupyter Notebook implementing the neural mass model reported in the paper.
empiricalfMRIAnalyses directory: MATLAB code used for empirical fMRI analyses. Start with masterscript.m. Please note that the code has not been tested for use on other servers, versions of MATLAB, etc. Feel free to contact the corresponding author with questions.
Notes on running FIR regression to correct task-state FC confounds:
- Many software packages are available for running FIR GLMs, such as FSL or AFNI
- In principle, any regression software can be used to run an FIR regression, so long as you have an FIR design matrix. In practice it is helpful to use a regression approach that can deal well with collinearity, such as a pseudo-inverse approach as is used in sklearn in Python (https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LinearRegression.html)
- Here is a function we created to help convert a (non-convolved) fMRI task timing design matrix into an FIR design matrix: https://github.com/ColeLab/TaskFCRemoveMeanActivity/blob/master/empiricalfMRIAnalyses/convertTaskTimingToFIRDesignMat.m