Image mapping to standard space

Overview

Image mapping to standard space is often required in reporting results from voxel-based analysis. This tutorial outlines the steps necessary to accomplish this task. I assume you have gone through the image registration and the image mapping tutorials and are familiar with the concepts and commands introduced in those tutorials.

Recommended standard-space DTI template

We recommend the Enhanced ICBM template according to the recent evaluation study "Investigating the role of ICBM-space human brain diffusion tensor templates in inter-subject spatial normalization" (pdf). The template is available for download here. The two files to use are IITmean_tensor_256.nii.gz, which is the tensor template, and IITmean_tensor_mask_256.nii.gz, which is the corresponding binary mask.

This sets the volume's dimensions (size) to powers of 2. We set the voxel dimensions (vsize) to [182/128 218/128 182/128] mm, so that we maintain the total physical size of 182x218x182 mm^3.

Mapping the Population Template to a Template Defining a Standard Space

Let the population template be mean.nii.gz and the template defining a standard space be template.nii.gz. The following set of commands will register the population template to the standard-space template with a series of registrations, beginning with rigid, then affine, then finally diffeomorphic.

dti_rigid_reg template.nii.gz mean.nii.gz EDS 4 4 4 0.001
dti_affine_reg template.nii.gz mean.nii.gz EDS 4 4 4 0.001 1
dti_diffeomorphic_reg template.nii.gz mean_aff.nii.gz template_mask.nii.gz 1 6 0.002

Recall that mean_aff.nii.gz is the result from the affine registration and template_mask.nii.gz is created with the following commands.

TVtool -in template.nii.gz -tr
BinaryThresholdImageFilter template_tr.nii.gz template_mask.nii.gz 0.01 100 1 0

Next, we compute the combined deformation field that defines the mapping from the population template space to the standard-space template.

dfRightComposeAffine -aff mean.aff -df mean_aff_diffeo.df.nii.gz -out mean_combined.df.nii.gz

Recall that mean.aff is the result from the affine registration, mean_aff_diffeo.df.nii.gz from the diffeomorphic registration, and mean_combined.df.nii.gz is our desired mapping.

Mapping Subject Native Space to the Standard Space

This step should be repeated for every subject in the studies. It involves the composition of the mapping from the subject native space to the population-template space and the mapping from the population template space to the standard space. Assume the original subject be subj.nii.gz, i.e., before any registration. Then the composition is accomplished with the following command:

dfComposition -df2 subj_combined.df.nii.gz -df1 mean_combined.df.nii.gz -out subj_to_template.df.nii.gz

Recall that subj_combined.df.nii.gz is the mapping from the subject native space to the population-template space.