MNI spaces

AFNI Message Board
1

I have aligned all my single subject data to the MNI152 space, which gives me much cleaner results–BOLD activity seems to conform more closely to grey matter, and leaks far less into ventricles and out of the brain. Previously I’d only ever used the N27 atlas.
I now want to do some ROI analysis based on anatomical definitions.

I have added the atlases recommended here: https://afni.nimh.nih.gov/afni/community/board/read.php?1,156135,156168#msg-156168 , and creating masks on these gives me a mask with the following dimensions:


R-to-L extent:   -74.000 [R] -to-    74.000 [L] -step-     2.000 mm [ 75 voxels]
A-to-P extent:   -71.500 [A] -to-   114.500 [P] -step-     2.000 mm [ 94 voxels]
I-to-S extent:   -62.500 [I] -to-    89.500 [S] -step-     2.000 mm [ 77 voxel

However 3dinfo on my MNI space single subect analysis gives me:


R-to-L extent:   -95.000 [R] -to-    95.000 [L] -step-     2.000 mm [ 96 voxels]
A-to-P extent:   -95.000 [A] -to-   131.000 [P] -step-     2.000 mm [114 voxels]
I-to-S extent:   -77.000 [I] -to-   113.000 [S] -step-     2.000 mm [ 96 voxels]

Which of these is the proper dimensions? Have I warped my single subject data incorrectly? The relevant line from the proc.py call, having passed the tlrc argument in the block list, is


-tlrc_base MNI152_T1_2009c+tlrc \

I’d be grateful for any help!

2

When you warp your data, you create a warp so that source dset SS is “well aligned” to base dset BB. When you apply that warp, you can choose whether to keep the source data approx. at its native resolution, or whether to resample it to match that of the base dset. To make the latter happen, you would apply the warp you have estimated and refer to the base dset as “-master BB”-- that should produce the output on the same grid as BB.

–pt

3

Previously, I have always used the tlrc option in afni_proc.py, and aligned to TT_N27. I can then create an ROI using the TT_N27 atlas, which, apart from a simple 3dresample -dxyz in order to get the right voxel size, has the right dimensions.

However, having now aligned to MNI152_T1_2009c, and resampled to the correct voxel size, the dimensions are off. The afni_proc.py script is exactly the same, other than having changed the -tlrc_base to a new atlas.

I previously aligned the same data to TT_N27, with the same script and the dimensions of the output and the atlas were the same.

So is there some step in the afni_proc.py I’m missing? Is there another option to add to get it to be on the same grid?

Given that the actual dimensions of my data are not the same as those of the atlas (i.e. the extent of the matrix is broader in every dimension), I don’t know whether simply 3dwarping will do the job. It seems like the dimensions need to be cut, not shrunk.

4

When I run 3dinfo on MNI152_T1_2009c+tlrc, I get these as dimensions on the file:


R-to-L extent:   -96.000 [R] -to-    96.000 [L] -step-     1.000 mm [193 voxels]
A-to-P extent:   -96.000 [A] -to-   132.000 [P] -step-     1.000 mm [229 voxels]
I-to-S extent:   -78.000 [I] -to-   114.000 [S] -step-     1.000 mm [193 voxels]

so, the FOV is 193 mm x 229 mm x 193 mm. If I had 2 mm isotropic voxels to fit in this same FOV, then I would have 96 x 114 x 96 of them. That seems consistent with what you have-- you fill the field of view of the “base” tlrc dataset with voxels of your specified size.

There’s a lot of files specified in the link you set, so I don’t know which you are referring to specifically; by N27, do you mean “TT_N27+tlrc”?

–pt

5

Thanks, I think I see where my error was, and it seems like I may be using an atlas with different dimensions?

I tried to use MNI_caez_ml_18+tlrc, atlas name CA_ML_18_MNI. It seems like this has different dimensions from the MNI152_T1_2009c+tlrc, which seems to be the primary MNI base dataset that comes with afni.

Do you know if there is an atlas available that has the macro labelling (similar to TT_N27+tlrc), but that fits the dimensions of the MNI152_T1_2009c+tlrc? Or do I need to realign all of my data to the CA_ML_18_MNI in order to use that atlas? All the other atlases that were downloaded from the link are probabilistic maps, rather than macro labelled.

6

The Eickhoff-Zilles atlases have different dimensions than the MNI templates. The ones provided by them with the anatomy toolbox match the N27 dataset with all in an MNI_ANAT space. The ones that have been shifted to MNI space are, in fact, in MNI space, so that x,y,z locations correspond to the MNI templates, but the grids, as you have noticed, are different. You can take each of the MNI versions of these atlases and make the grid match with 3dZeropad as shown below.

3dZeropad -master ~/abin/MNI152_T1_2009c+tlrc. -prefix MNI_caez_mpm_18_zp ~/MNI_atlases/MNI_caez_mpm_18+tlrc.
3drefit -copytables ~/MNI_atlases/MNI_caez_mpm_18+tlrc. MNI_caez_mpm_18_zp+tlrc.
3drefit -cmap INT_CMAP MNI_caez_mpm_18_zp+tlrc.

7

That did the trick, thanks very much for all the help!