Getting functional atlas masks from MNI to native space

, ,


I am approaching tractography using HCP DTI dataset. I would like to use as a seed points WMGMI of visual cortex as defined in the Juelich Histological atlas. However, atlas is in MNI space and my scans (DTI, T1, T2 and so on) are in scanner native space.

I am aware of transformations between spaces. Hence, I was trying to transform mask for functional region from MNI to native space.
For that I used flirt from FSL to produce a transformation matrix going from MNI image → my native space (my native space as a reference). Then I tried to apply this transformation by using mask from Juelich atlas as an input, my T1 as a reference and additional parameters -applyxfm -init [TRANSFORM MATRIX FROM PREV. STEP]. However if I open my T1 in e.g. 3DSlicer, location of the mask is off (wrong lobe, hemisphere etc.).

What am I doing wrong here? Am I forced to transform all of my images to MNI space? If I do so with my T1 (brain extracted with BET2), the image looks really odd → rotated planes, different angles etc.


What images did you use to inform your initial transformation?


In the first approach I used MNI152_1mm_brain from fsl as input, and my subject T1 as reference.

Now I am testing a “warping” approach found here: Use of volumetric templates & parcellation atlases - Wiki - MRtrix3 Community

No idea why I haven’t found that tutorial earlier…