Multiscale parcellation MNI reference

211 views
Skip to first unread message

Pavan Chaggar

unread,
May 1, 2020, 7:05:55 AM5/1/20
to cmtk-users
Hello, 

I am looking to retrieve parcellations in standard MNI space, is this possible in the current implementation of the multi-scale brain parcellator? Or, are you aware of/have any standard reference parcellations that are available for download (I have not been able to find one)? I am hoping to map MNI coordinates for labelled areas to nodes of a connectome that is based on the Lausanne 2008 parcellation. 

Thanks in advance, 
Pavan

Sebastien Tourbier

unread,
May 5, 2020, 10:04:07 AM5/5/20
to cmtk-users
Hi Pavan,

Up to now, there is no such parcellations in standard MNI space.

You can still find the freesurfer annot files in freesurfer fsaverage space at:

The solution I see up to now to get the parcellation in MNI space would be:
(1) to use a T1w image in MNI space (for instance Colin27 dataset - https://www.mcgill.ca/bic/software/tools-data-analysis/anatomical-mri/atlases/colin-27) that you organized as a sample BIDS dataset
(2)to process this sample Colin27 BIDS dataset via the multi-scale brain parcellator. The resulting parcellations will be in MNI space.
(3) Then, if you want to get the coordinates of center of mass of 3D parcellation ROI you could use for instance nilearn function find_parcellation_cut_coords (https://nilearn.github.io/modules/generated/nilearn.plotting.find_parcellation_cut_coords.html#nilearn.plotting.find_parcellation_cut_coords)

Hope this helps,
Sebastien

Pavan Chaggar

unread,
May 12, 2020, 7:48:52 AM5/12/20
to cmtk-users
Hi Sebastien, 

Thanks for your reply. 
I have tried this with some success. Do the parcellation results vary between images or should one always receive the same parcellation? 
The reason I ask is that I am trying to match the parcellation regions with those in the Budapest reference connectome (https://braingraph.org/cms/download-pit-group-connectomes/), who also use cmtk for their connectome generation pipeline, however, for scales 2:5 there are differences in the labelled regions. There are some regions which are present in the connectome but not my parcellation of an MNI brain, and vice versa. For example, the connectome container lh-entorhinal-3, whereas my parellation contains only lh-entorhinal1 and 2. There are a relatively small number of regions missing (between 6-30 regions depending on scale) and the differences seem to be only the left hemisphere. Additionally, it seems the problems are with the denomination of subregions, e.g. 3 lg-entorhinal vs 2; that is, there aren't any regions missing, just different sub-regions. The parcellation and connectome match exactly for scale1. 

Best wishes, 
Pavan

Pavan Chaggar

unread,
May 12, 2020, 8:24:52 AM5/12/20
to cmtk-users
I forgot to mention, for scaled 3:5 there are also a different number of nodes. I've summarised the differences below: 

Scale 2

number of regions in Budapest connectome: 129 
number of regions in mni parcellation: 129 

Regions in Budapest connectome not in parcellation 

['Brain-Stem' 'lh-postcentral-3' 'lh-precentral-4'
 'lh-rostralmiddlefrontal-3']

Regions in parcelletion not in Budapest connectome 

['brainstem' 'lh-inferiorparietal-3' 'lh-lateraloccipital-3'
 'lh-medialorbitofrontal-2']


Scale 3

number of regions in Budapest connectome: 234
number of regions in mni parcellation: 231

Regions in Budapest connectome not in parcellation 

['Brain-Stem' 'lh-bankssts-2' 'lh-insula-4' 'lh-lingual-4'
 'lh-postcentral-6' 'lh-postcentral-7' 'lh-precentral-7' 'lh-precentral-8'
 'lh-superiorfrontal-9' 'lh-supramarginal-5']

Regions in parcelletion not in Budapest connectome 

['brainstem' 'lh-cuneus-2' 'lh-inferiorparietal-6'
 'lh-medialorbitofrontal-3' 'lh-paracentral-3' 'lh-parstriangularis-2'
 'lh-pericalcarine-2']



Scale 4

number of regions in Budapest connectome: 463 
number of regions in mni parcellation: 461

Regions in Budapest connectome not in parcellation 

['Brain-Stem' 'lh-caudalmiddlefrontal-6' 'lh-inferiortemporal-8'
 'lh-isthmuscingulate-3' 'lh-lateraloccipital-11' 'lh-lingual-8'
 'lh-postcentral-13' 'lh-postcentral-14' 'lh-precuneus-11'
 'lh-superiorfrontal-18' 'lh-superiorparietal-14' 'lh-supramarginal-10'
 'lh-transversetemporal-2']

Regions in parcelletion not in Budapest connectome 

['brainstem' 'lh-caudalanteriorcingulate-3' 'lh-cuneus-4'
 'lh-inferiorparietal-11' 'lh-inferiorparietal-12' 'lh-middletemporal-8'
 'lh-middletemporal-9' 'lh-paracentral-6' 'lh-parstriangularis-4'
 'lh-pericalcarine-4' 'lh-rostralmiddlefrontal-13']


Scale 5

number of regions in Budapest connectome: 1015
number of regions in mni parcellation: 1017

Regions in Budapest connectome not in parcellation 

['Brain-Stem' 'lh-caudalmiddlefrontal-12' 'lh-caudalmiddlefrontal-13'
 'lh-entorhinal-3' 'lh-fusiform-18' 'lh-insula-17' 'lh-isthmuscingulate-7'
 'lh-parsopercularis-10' 'lh-postcentral-31'
 'lh-rostralanteriorcingulate-5' 'lh-superiorfrontal-43'
 'lh-superiorfrontal-44' 'lh-superiorfrontal-45' 'lh-superiortemporal-26'
 'lh-supramarginal-21' 'lh-transversetemporal-4']

Regions in parcelletion not in Budapest connectome 

['brainstem' 'lh-caudalanteriorcingulate-6' 'lh-cuneus-8'
 'lh-inferiorparietal-23' 'lh-inferiorparietal-24'
 'lh-inferiorparietal-25' 'lh-inferiorparietal-26'
 'lh-lateralorbitofrontal-17' 'lh-medialorbitofrontal-11'
 'lh-middletemporal-17' 'lh-middletemporal-18' 'lh-middletemporal-19'
 'lh-paracentral-12' 'lh-parstriangularis-8' 'lh-pericalcarine-8'
 'lh-precentral-36' 'lh-precuneus-23' 'lh-rostralmiddlefrontal-27']
Reply all
Reply to author
Forward
0 new messages