Granger causality analysis implementation on MATLAB: A graphic user interface toolkit for fMRI data processing

A lot of functional magnetic resonance imaging (fMRI) studies have indicated that Granger causality analysis (GCA) is a suitable method to reveal causal effect among brain regions. Based on another MATLAB GUI toolkit, Resting State fMRI Data Analysis Toolkit (REST), we implemented GCA on MATLAB as a graphical user interface (GUI) toolkit. This toolkit, namely REST-GCA, could output both the residual-based F and the signed-path coefficient. REST-GCA also intergrates a programme that could transform the distribution of residual-based F to approximately normal distribution and then permit parametric statistical inference at group level. Using REST-GCA, we tested the causal effect of the right frontal-insular cortex (rFIC) onto each voxel in the whole brain, and vice versa, each voxel in the whole brain on the rFIC, in a voxel-wise way in a resting-state fMRI dataset from 30 healthy college students. Using Jarque-Bera goodness-of-fit test and the Lilliefors goodness-of-fit test, we found that the transformation from F to F′ and the further standardization from F′ to Z score substantially improved the normality. The results of one sample t-tests on Z score showed bi-directional positive causal effect between rFIC and the dorsal anterior cingulate cortex (dACC). One sample t-tests on the signed-path coefficients showed positive causal effect from rFIC to dACC but negative from dACC to rFIC. All these results indicate that REST-GCA may be useful toolkit for caudal analysis of fMRI data.

► We implemented a free MATLAB GUI toolkit, REST-GCA.
► We tested REST-GCA using a set of resting-state fMRI data and found a consistent result with that of previous studies.
► For the first time, we transformed the residual based F to approximately normally distributed Z score.
► This procedure substantially improved the normality and hence allow parametric statistical analysis.