Average T1w or T2w repetitions Bruker 7T - GarzaLab/Documentation GitHub Wiki

Because no 2 images are alike, you need to first register them as good as possible before averaging brains.

You need to have installed:

1. Ants

This is an example from a T1w FLASH for DBM.

New version

Gabriel Devenyi from the CobraLab supports this solution.

Create a new antsMultivariateTemplateConstruction2_rigidaverage.sh from the original.

The new version will have the following modifications:

1. Remove the Sharpen command on 275
2. Write --interpolation BSpline[5] to every ${WARP} line

Or just download my version from here. You can copy it in your ANTs directory for easiness of use.

The script that will take care of this is located here.

Files were prepared with brkraw and the structure should be as such:

data
├── sub-10
│   ├── ses-1
│   │   ├── anat
│   │   │   ├── sub-10_ses-1_T1w.json
│   │   │   ├── sub-10_ses-1_T1w.nii.gz
│   │   │   ├── sub-10_ses-1_T2w.json
│   │   │   └── sub-10_ses-1_T2w.nii.gz
│   │   ├── fmap
│   │   │   ├── sub-10_ses-1_fieldmap.json
│   │   │   ├── sub-10_ses-1_fieldmap.nii.gz
│   │   │   └── sub-10_ses-1_magnitude.nii.gz
│   │   └── func
│   │       ├── sub-10_ses-1_task-rest_run-1_bold.json
│   │       └── sub-10_ses-1_task-rest_run-1_bold.nii.gz
│   └── ses-2
│       ├── anat
│       │   ├── sub-10_ses-2_T1w.json
│       │   ├── sub-10_ses-2_T1w.nii.gz
│       │   ├── sub-10_ses-2_T2w.json
│       │   └── sub-10_ses-2_T2w.nii.gz
│       ├── fmap
│       │   ├── sub-10_ses-2_fieldmap.json
│       │   ├── sub-10_ses-2_fieldmap.nii.gz
│       │   └── sub-10_ses-2_magnitude.nii.gz
│       └── func
│           ├── sub-10_ses-2_task-rest_run-1_bold.json
│           └── sub-10_ses-2_task-rest_run-1_bold.nii.gz

Run the script inside your project folder with data in a similar structure as the tree above. In this example I'm using the data shown above:

bash rat-invivo-average.sh sub-5 ses-1 T1w

Outputs:

The script will create a folder called derivatives/sub-5/ses-1/preproc with the following important files:

average.mnc
average.nii.gz
average.jpg

The average.mnc file goes to the next preprocessing step. Check the Preprocessing.

Old version

Transform the nifti (Bruker2Nifti) to MNC. Separate the time using fslsplit, then use ants.

fslsplit rat.nii.gz -t

nii2mnc rat.nii.gz rat.mnc

antsRegistration --dimensionality 3 --float 0 --output [reg_1,rat1_reg.mnc] --interpolation Linear --use-histogram-matching 0 --initial-moving-transform [rat1.mnc,rat2.mnc,1] --transform Rigid[0.1] --metric MI[rat1.mnc,rat2.mnc,1,64,None] --convergence [2000x1000x500x250x100,1e-6,10] --shrink-factors 4x3x2x1x1 -s 2x1.5x1x0.5x0vox -v

Ants changes the dimensions, so we need to change that.

mincresample -like rat2.mnc rat1_reg.mnc rat1_reg_dim.mnc

Then you average

mincaverage rat1.mnc rat2.mnc rat_out.mnc