Creating surface dataset from .gii files

[samantha audrain]

Hi there,

 

I am trying to create a surface dataset from a .gii file. However, cosmo_surface_dataset isn't compatible with .gii files. Is there a way to get around this?

 

I can read in the .gii file using the gifti function in matlab but I am unsure of how to format it as a surf_ds. My surface files were created with HippUnfold, which uses .gii files. 

Is it possible to do this or am I out of luck?

 

Thanks!

 

Sam 

[Nick Oosterhof]

Greetings,

On 25 Apr 2024, at 19:40, samantha audrain <samanth...@gmail.com> wrote:

I am trying to create a surface dataset from a .gii file. However, cosmo_surface_dataset isn't compatible with .gii files. Is there a way to get around this?

 

I can read in the .gii file using the gifti function in matlab but I am unsure of how to format it as a surf_ds. My surface files were created with HippUnfold, which uses .gii files. 

There is actually support for gifti files using cosmo_surface_datset, using this library: https://github.com/gllmflndn/gifti. 

(I see now the documentation did not mention the option, I just added it).

Have you tried it?

[samantha audrain]

Hello and thanks for your response! 

Can you clarify regarding the support available for gifti files using cosmo_surface_dataset? I am aware of the gifti library and I can read my gifti in to matlab using it. But how do I then get cosmo_surface_dataset to turn it into the surf_ds format?  When I feed the cosmo_surface_dataset function the imported gifti file, it gives me the following error:

Error using cosmo_surface_dataset>find_single_or_error
unknown object of type gifti

Or is there some other way I should be using the gifti library you mentioned?

[Nick Oosterhof]

On 26 Apr 2024, at 16:41, samantha audrain <samanth...@gmail.com> wrote:

Can you clarify regarding the support available for gifti files using cosmo_surface_dataset? I am aware of the gifti library and I can read my gifti in to matlab using it. But how do I then get cosmo_surface_dataset to turn it into the surf_ds format?  When I feed the cosmo_surface_dataset function the imported gifti file, it gives me the following error:

Error using cosmo_surface_dataset>find_single_or_error
unknown object of type gifti

Or is there some other way I should be using the gifti library you mentioned?

Currently cosmo_surface_dataset detects a gifti object ``g`` through:

- g must be a gifti object (from the gifti library), that is isa(g, ‘gifti’) must be struct

- g must have a field cdata. At the moment this function was written that was what I observed.

Is the structure of your gifti object different? What does it look like?

[samantha audrain]

It's a gifti object but isn't a struct, isa(g, 'struct') returns 0. And there is no cdata field. 

intermediate_fn=fullfile(surface_path,...

sprintf('sub-pilot5_hemi-L_space-T2w_den-0p5mm_label-hipp_midthickness.surf.gii'));

g=gifti(intermediate_fn)

[Nick Oosterhof]

On 30 Apr 2024, at 15:53, samantha audrain <samanth...@gmail.com> wrote:

It's a gifti object but isn't a struct, isa(g, 'struct') returns 0. And there is no cdata field. 

intermediate_fn=fullfile(surface_path,...

sprintf('sub-pilot5_hemi-L_space-T2w_den-0p5mm_label-hipp_midthickness.surf.gii'));

g=gifti(intermediate_fn)

gifti comes in many forms. For CoSMoMVPA two are relevant:

- anatomical datasets with faces and vertices. These can be passed in as the (second) surfs argument in cosmo_surficial_neighborhood. Your .surf.gii file would fall in this category. (Note that CoSMoMVPA does not use the .mat field). 

- functional datasets, with data for all or a subset of the nodes. These may have the .func.gii extension, although that is just a convention. Such files can be used as input to cosmo_surficial_dataset, and also as the ds (first) argument in cosmo_surficial_neighborhood. 

For searchlights, one could also select functional data in the volume while still using anatomical surface dataset, also using cosmo_surficial_dataset - this may preserve more information leading to more sensitivity. This approach was used in (shameless self-promotion) https://www.diedrichsenlab.org/pubs/Oosterhof_Neuroimage_2011.pdf

[samantha audrain]

Ah I see now, I was confused by the terminology in my newbie state. All is working now, thanks for pointing me in the right direction! And for the paper.