Sulcal label

[jjiang1263]

Hello，
I would like to ask about the "Create Sulcus Label Volume" tool. How should the input_int_to_label_translation file be obtained? Also, in which space is the output volume data located, and how can it be aligned to the MNI standard space?

Thanks，
Lewis

[denisri]

Hi,

input_int_to_label_translation is an optional parameter and should not always be specified. It is basically created by a first call to Create Sulcus Label Volume (as the int_to_label_translation parameter) and can be then reused in later calls to the process in order to preserve a consistent common regions numbering. This, thus, prevents to use the process in a "direct" iteration because the input_int_to_label_translation parameter should be empty for the first iteration, and filled for the next ones. If you need to do it, either call it a first time to create the file, then use it in all iterations. Or maybe you can input an empty file (I'm not sure of that, I haven't tried it before answering).
You can align the output data by providing transformation_matrix and setting transformation to Yes. The transformation matrix will typically be the Raw T1 to Talairach-MNI template transformation (.trm found in the Brainvisa Morphologist outputs data), and you may specify a transformation_template image to set the output image size (it can be the MNI template image).
I admit this process is rather old and could/should be reworked for simplicity of use...

Denis

[jjiang1263]

Thank you very much!

Lewis

[jjiang1263]

Hello, sorry to bother you again. I obtained the volume label data using the tool and noticed that many of the sulcus labels are located in the cerebrospinal fluid area when combined with the brain image. How can I extract the data on the cortex?

[denisri]

Yes that's it ! Well, actually, it depends on what you mean by "sulcus". For us, it's the fold itself, that is, the skeleton of the CSF going inside the fold, so yes, for us it's CSF, and it's the way we use it. Now I understand you have a different definition of sulcus. In the outputs of the Morphologist pipeline, there is an image named "roots" (Lroots_<subject>.nii.gz in the segmentation directory). This image is a parcellation of the cortex based on the topological entities which define the sulcal graph vertices (obtained via the "skeleton" image). Each value here corresponds to a vertex in the sulcal graph (L<subject>.arg). The correspondance is found in the graph vertices, each vertex has a rootsbassin attribute which is the number of the region in the roots image. Is that clear ?
To manipulate the cortical graph, there are python libraries shipped with BrainVisa.
However once the sulci are identified, additional splits may occur: a graph vertex may be split in two or more pieces which will get different sulcus names. In that case several vertices may refer to the same rootbassin and there is no obvious method to split those regions according to sulci labels (we could probably perform a voronoi diagram inside the bassin but it's not done).
Denis

[jjiang1263]

I understand,thank you very much!
Lewis