search

ANTsX / ANTs

Advanced Normalization Tools (ANTs)
Apache License 2.0
1.21k stars 381 forks source link

How to build template for incomplete data? #695

Closed bkossows closed 5 years ago

bkossows commented 5 years ago

Hi, I've got DTI FA maps of cat brains. The data is longitudinal and I tried building the template from 5 timepoints of each individual with mvtc2.sh. The data is acquired with full coverage in case of 3 timepoints and reduced number of slices in 2 of them (top row). I found that the images easily register too much with deformable transforms as well as affine only (bottom row). Please let me know how to deal with such images. Exemplary template was build with /opt/software/ants_scripts/antsMultivariateTemplateConstruction2.sh -d 3 -c 2 -i 6 -j 2 -m CC[4] -k 1 -q 100x70x30x3 -f 8x4x2x1 -s 3x2x1x0 -n 1 -r 1 -l 1 -t Affine -y 1 *_eddy_dti_FA.nii github_mvtc_ra_001

cookpa commented 5 years ago

You might be able to improve registration with a mask:

https://github.com/ntustison/PartialSlabEpiT1ImageRegistration

As for your template, it's going to be tough to build an unbiased average with a large chunk of the brain missing for two of the time points. You might be able to build a template from the complete coverage images with the template script and then register the partials to that.

cookpa commented 5 years ago

Also, I don't know if you've cropped your screen shots, but if your image extent is tight to the edge of the brain, adding empty space around the image with the ImageMath PadImage command will likely help.