gamRandomise: Generate FSL Randomise call for a GAM Model

In voxel: Mass-Univariate Voxelwise Analysis of Medical Imaging Data

Description

This function is able to generate all the necessary files to run randomise with a GAM Model This script will write out all design and contrast files This function will run a f-test to compare a full and reduced model (a model with and without spline)

Usage

gamRandomise(image, maskPath = NULL, formulaFull, formulaRed, subjData,
  outDir, nsim = 500, thresh = 0.01, run = FALSE)

Arguments

image	Input path of 'nifti' image or vector of path(s) to images. If multiple paths, the script will all mergeNiftis() and merge across time.
maskPath	to mask. Must be a binary mask
formulaFull	Must be the formula of the full model (i.e. "~s(age,k=5)+sex+mprage_antsCT_vol_TBV")
formulaRed	Must be the formula of the reduced model (i.e. "~sex+mprage_antsCT_vol_TBV")
subjData	Dataframe containing all the covariates used for the analysis
outDir	output directory for randomise
nsim	Number of simulations
thresh	significance threshold
run	FALSE will only print randomise command but won't it

Value

Return randomise command

Examples

## Not run: 

subjData = mgcv::gamSim(1,n=400,dist="normal",scale=2)
OutDirRoot="Output Directory"
maskName="Path to mask"
imagePath="Path to output"
covsFormula="~s(age,k=5)+sex+mprage_antsCT_vol_TBV"
redFormula="~sex+mprage_antsCT_vol_TBV"

gamRandomise(image = imagePath, maskPath = maskName, formulaFull = covsFormula, 
           formulaRed = redFormula, subjData = subjData, outDir = OutDirRoot)


## End(Not run)

voxel documentation built on May 1, 2019, 10:26 p.m.