Export specific sulcus as both ROI (nii.gz) and mesh (vtk)

[michael]

Hello,

I would like ideally to batch process the segmentation and exportation of a specific sulcus as both a ROI and a vtk file. That's my current goal and any tip on this process would be very welcome.

Right now i'm stuck on two small problems and the main priority is getting the ROI file for this sulcus:

1) When i visualize the sulcus i'm interested in (for example with Racctom_subject_default_session_auto.arg in the right hemisphere) it is broken into several nodes (name: unknown(342), unknown(187), unknown(291), unknown(39), ...), so how to find all nodes with the same sulcus label and how to save them in a single ROI? (and in a single vtk mesh?)
(PS: is it normal that all names as set to "unknown"? Ideally it would be great if i could have them including the label name, such as: my_sulcus(1), my_sulcus(2), my_other_sulcus(1), etc..)

2) When i save one of this node to an ROI (in Anatomist, right click on the node then File --> Export mask as volume), the resulting volume dimensions doesn't match the other processed volumes (including T1). I guess it means the resulting mask is in the reference space in which the sulci recognition has been carried out, right? Then how do i get the sulcus segmentation back in the subject space in which i have all my other volumes? Should i be worried about the differences between the resulting segmented sulcus (once "inverse transformed") compared to its position in subject space (such as in the skeleton_acctom volume)?
(PS: what is the reference space in which the sulci recognition is done?)

Many thanks for your help,

Michael

[michael]

I finally succeeded in obtaining the sulcus in a ROI. However i am still stuck on obtaining the associated vtk mesh and on serializing this process for several subjects so any help on this two issues would be very welcome.

[riviere]

Hi Michael,
The ROIs exported from the graph have the bounding box of the graph itself (as far as I remember) so are smaller than the original image, but it merely corresponds to a shift in native space, not to a normalized space.
See this post for a related question.
To get the meshes, you can either extract them from nodes comosing the same sulcus as in the above post, or remesh the exported ROIs (using AimsMesh or AimsMeshBrain commands).

Denis