Fit Talairach atlas in MNI space using Nilearn


I am trying to create a mask of the dlPFC, using the Talairach atlas.

However, the rest of my analysis, and the image I am working on, are in MNI space

image = 'Data/smooth_clean_sub-210409dc_task-rest_space-MNI152NLin6Asym_res-2_desc-preproc_bold.nii'
niimg = datasets.load_mni152_template()

I import my atlas as such :

atlas = nilearn.datasets.fetch_atlas_talairach("ba", data_dir=None, verbose=1)
atlas_img = atlas.maps
# Extract the dlPFC
dlPFC_Brodmann = mean_img([math_img('img == %d'  %i, img=atlas_img) for i in [43,64,70]])

    cut_coords=(-41, 43, 27),

When I plot atlas_img, one can notice that the atlas does not fit my MNI template (too small):

Can one help me to fit this atlas exactly on my template? I tried to multiply the image affines, using flirt in FSL, but failed to get good results.

Thank you very much for your help!

The problem you are trying to deal with is that there is no MRI template or other type of template available that matches the Talairach daemon. That means the atlas can’t be used with any atlas accurately including ones even already in Talairach space. The atlas itself is too limited. The dorsolateral prefrontal cortex has had multiple definitions over the years as combinations of Brodmann areas 46 and 9. Those have a great many definitions for each of those that look very different from one atlas to the other. The JulichBrain defines several new smaller regions and a “gap” region to make the dorsolateral prefrontal cortex region, described in the link below. It’s available with AFNI or through the EBRAINS website.

1 Like