Spatial Statistical Analysis Tools

Spatial parametric tools

SimpleSpatialModel

Class for fitting of linear models to spatial distribution of voxels using the weighted least squares method. Full class documentation is provided here.

Model and correct bias field: imBias

Uses SimpleSpatialModel to estimate and correct bias field.

imBias image.gipl [options]

Options include:

-m mask.gipl weight/mask image

-d degree maximum degree of polynomial model

-b bias.gipl save bias field

-c corrected.gipl save corrected image

Voxel parametric tools

SimpleSPM

A tool to carryout Statistical Patametric Mapping (SPM) -like voxelwise linear modelling and  statistical significance testing on sets of images in a multi-subject imaging study.This is carried out without the need for SPM or Matlab and is implemented efficiently in C++ using openmp. It also allows correction for multiple comparisons using a parallel implementation of Permutation correction.

This command line program accepts a configuration file as an argument and a set of output files to store the result:

SimpleSPM inputConfig.txt OutputModelParamsB.gipl OutputTstatEst.gipl [options]

Two basic output files specified are a 4D file of the model parameters estimated at each voxel and the T statistics for the fit of each model parameter to the data.

Options include:

-DataMap DataDump.gipl Optional Dump of All input images as one 4D gipl file.

The configuration file specifies the number of parameters andthe input data files along with a brain mask and input data value scaling. Subject specific model parameters can be specified directly in this file or specified as being stored in subject specific XML files. A simple example is shown below:

=============================================

%!SimpleSPMSetup1.4
NumSubj:   6

NumVar:        2
RefMask: BrainRegion.gipl

Image: ChangeMap.gipl

ImageScaleOffset: 1000.0 0.0 0
ImageInfo: *.xml
NumPerm: 3000 PermDistUNDC.txt
../ADsubject1   1  <subjectAge>  1
../ADsubject2   1  <subjectAge>  1
../ADsubject3   1  <subjectAge>  1
../CNsubject4   0  <subjectAge>  1
../CNsubject5   0  <subjectAge>  1
../CNsubject6   0  <subjectAge>  1

top