Creating the fs_default.txt file for a new parcellation scheme

Hi everyone,

I was wondering whether anyone can lead me to a tutorial on how to create the fs_default.txt for an arbitrary atlas instead of using the Desikan atles. I want to parcellate my data into a higher number of regions and I a not sure how to create the LUT using the new atlas I want to use.

Really appreciate all of your help!

Regards,
Pubu

Hi Pubu,

I haven’t written a “tutorial” per se on creation of lookup tables. I have at least written an additional tutorial page on use of the labelconvert command, which hopefully demonstrates the purpose of such lookup tables and how their data are utilised internally within the code. In terms of formatting, even without an understanding of C++ the code here gives a reasonable insight into the range of possible formats that the lookup table can take (basically it supports all of the raw-text formats I have encountered in the wild thus far); but chances are if you’re looking to obtain a more fine parcellation within FreeSurfer, you won’t need to worry about the format of the table itself: you just need to modify the contents of the existing examples to suit that finer parcellation.

If you still have outstanding questions not solved by those two links, please let me know. An entirely generic tutorial would probably be difficult to write given the range of possible mechanisms by which a parcellation could be derived, but maybe there’s some points missing in that documentation page (or maybe it just needs renaming).

Rob

Hi Rob,

Thank you for your response. I had gone through those links. But I couldn’t find what I am looking for in either of those pages.
I got the fs_default.txt file for 84 parcellations from the Human Connectome Project. So I think I kind of have an idea about the format of the file and also the usage of it. Now I would like to try a different parcellation scheme with more than 84 regions (208 to be specific). I have the parcellation mask for this new parcellation I’d like to use. But I am looking for a way to get the fs_default.txt file for that parcellation.

Regards,
Pubu

Pubu,

As best I can understand, you have a label image for a non-standard parcellation, which you wish to use in connectome construction; but you require a lookup table file that will convert that parcellation image into integer values that increment from 1 upwards. There’s no command that will generate this file for you; you need to have in your possession an existing lookup table on which the parcellation image in its current state is based, and manually create the target lookup table text file, which defines the order and numbering that you want the nodes to be mapped to when the parcellation image is fed through the labelconvert command.

Without a more specific description of either a practical issue or a limitation in understanding, I’m not sure that I can provide any more guidance over and above the labelconvert tutorial page regarding how this lookup table text file needs to be derived.

Rob

Hi Rob,

“As best I can understand, you have a label image for a non-standard parcellation, which you wish to use in connectome construction; but you require a lookup table file that will convert that parcellation image into integer values that increment from 1 upwards.”

Yes, you are correct! Then I might have to go back to the source and get that lookup table together with the label image. Thank you very much for clarifying things for me. I am sorry for asking a dumb question, I am still new to MRTRIX and trying to play around with it to learn.

Thank you again!

Pubu

Hello Pub and Rob,

This conversation was helpful. I too trying to use other parcellation atlas (Shen et al, 2015 from here: https://www.nitrc.org/frs/?group_id=51) to build the connectivity matrix. I have the Shen artlas in .nii image format.
I am not able to find out how to provide the input to the “labelconver” function. I mean how to create the fs_default.txt /look up table for the Shen atlas. Please suggest.

Thank you in advance
Raja

Above and beyond the labelconvert documentation page linked above, I can’t really advise on how to generate the necessary lookup table files without going to more effort in describing it than what would have been required to simply generate the file. I can only suggest reading that page multiple times and understanding it as best you can. You could also try modifying one of the files in share/mrtrix3/labelconvert/ (e.g. delete nodes, change values), re-running labelconvert, observing what effect that has on the output image, and see if that helps to make the operation of this command “click”.

Hi Raja,

Were you able to find a solution to your problem above? if yes, would you be able to share it here?

Thanks,
Mahmoud