It is better to have the termination surface outside the counting surface (e.g. count on white, terminate on pial). I do not remember if they can be the same surface or not, so you would need to ask this on the FSL list.
Peace, Matt. From: Timothy Coalson <[email protected]<mailto:[email protected]>> Date: Monday, April 16, 2018 at 4:10 PM To: Xinyang Liu <[email protected]<mailto:[email protected]>> Cc: HCP 讨论组 <[email protected]<mailto:[email protected]>>, Matt Glasser <[email protected]<mailto:[email protected]>> Subject: Re: [HCP-Users] using surfaces as termination masks Inline replies. Tim On Sun, Apr 15, 2018 at 9:42 AM, Xinyang Liu <[email protected]<mailto:[email protected]>> wrote: Dear Matt, dear Tim, Very grateful for both of your kind replies. :) Here are some further questions based on your feedbacks. 1. For the ROI of neocortex (atlasroi.shape.gii), does it mean this shape file includes pial, midthickness and white surfaces? I previously thought the .shape.gii file provides cortical shape information, like curvature. But then I found the value of each atlasroi vertex is 1. Is it just an atlas map? Why was it chosen to set the value of L/R white.asc in surf2surf for brain surfaces? How do we usually give the --values in surf2surf? .shape.gii is a file extension, it does not tell you what is in the file. It is actually the same file format as .func.gii, and we are not entirely consistent about when we choose to use each one. "atlasroi.shape.gii" is just a binary roi, as "atlasroi" is intended to imply - it has values of 1 on vertices that get used in our standard cifti space, and 0 on vertices that don't get used. There are no coordinates or topology in it, no white, no pial, no midthickness - it is not a .surf.gii file, and we don't use the "combined data and coordinates" type files that other software does (unfortunately, probtrackx requires such "combined" files, adding to your confusion). In workbench, surfaces are always loaded from a different file than the data to display on them. 2. For the CIFTI and GIFTI matrices, I didn't intend to get the value of each element, but preferred to know what kind of data is stored, which might be shown by the name of each column, or data can be recognized as coordinates. I am wondering how to check this. They can store any type of measure. You can get the names of the maps, if they have been named at all, by using "wb_command -file-information -only-map-names <file>". Cifti files only store data values, and do not additionally contain surface coordinates or topology. You have to load surfaces separately from the cifti file. 3. I could not understand well that the cortex ROI would allow connections to pass through the medial wall. Because when looking at the surface outlines of "L/R.white.32k_fs_LR.surf.gii", it seems that the left and white surfaces are closed regions (please see the figure attached). Then how do connections pass through the middle surfaces? The atlasroi.shape.gii has values of 0 on the vertices that cut through the corpus callosum - if you generate the files correctly, the tractography software will see these values of 0 and know not to terminate a streamline that hits the surface where the value is 0, and continue until it hits a vertex with a value of 1. You will need to specify the files for both hemispheres to the tractography command, I assume. 4. If we acquire seed ROIs from the midthickness surface, does it mean that the termination surface masks can only come from the pial surface then? Because the white surface is in the inner layer and therefore would stop the streamlines coming from the mid-surface to go inside the brain. Is my understanding correct? I don't understand this question. I don't know if it is even possible to specify multiple stopping surfaces for one hemisphere in probtrackx. If it were, and you used the white surface as one of them, then it should stop all streamlines before they reach midthickness or pial (unless you did something wrong). I think the general guidance is to use the white matter surface, as fiber directions become less reliable in gray matter. Look forward to your further guidance. Many thanks! Best regards, Xinyang At 2018-04-14 05:42:36, "Timothy Coalson" <[email protected]<mailto:[email protected]>> wrote: Inline comment. Tim On Thu, Apr 12, 2018 at 11:01 PM, Xinyang Liu <[email protected]<mailto:[email protected]>> wrote: 2. How to look at the CIFTI and GIFTI files as matrices? I haven't found a proper way to look at the information inside as rows and columns. Gifti files, and dense cifti files, have too many elements to be properly viewed as a matrix, and the vertices that are adjacent in the file data don't make much sense spatially. If you want a matrix display, you should use a parcellated cifti file (which can be made from a dense cifti file with wb_command -cifti-parcellate). Tim 1. That is the ROI of neocortex (excluding the medial wall). The medial wall is the non-cortical tissue that connects the two hemispheres (e.g. corpus callosum, diencephalon). All of the surface meshes are usable if they are the same kind of mesh (i.e. native, 32k, 164k). 2. You can convert them to dconn files with wb_command -probtrackx-dot-convert. 3. The cortex ROI will allow connections to pass through the medial wall. You can include multiple surface and volume ROIs in probtrackx. Peace, Matt. _______________________________________________ HCP-Users mailing list [email protected] http://lists.humanconnectome.org/mailman/listinfo/hcp-users
