Back to PLS HelpPLS of cortical thickness
I'm Online
Posted on 05/26/09 15:45:00
Number of posts: 8
Hi All,
I'd like to do a behavioral PLS on native cortical thickness (CT) measures. My CT values are initially stored in one text file for each subject, in which each line of the file has a thickness value (floating point; in millimeters). As our cortical surface is comprised of 81924 vertices, and since thickness is measured at each vertex, we have 81924 CT values per subject.
Now, as PLS wants nifti volumes in, I can translate vertex coordinates to voxel coordinates and move the CT values into a volume ... which is a bit of a kludge, but will work. Here are my questions: (1) does moving the CT values to a volume sound reasonable, or is there a better way?, and (2) once I have the native CT values in the volume, which PLS analysis option should I use: PET? VBM? I don't believe I want the CT values normalized in case overall thickness is associated with some of my behavioral measures ... if that makes a difference. I would be greatful for any suggestions.
Thanks,
-Jim
I'd like to do a behavioral PLS on native cortical thickness (CT) measures. My CT values are initially stored in one text file for each subject, in which each line of the file has a thickness value (floating point; in millimeters). As our cortical surface is comprised of 81924 vertices, and since thickness is measured at each vertex, we have 81924 CT values per subject.
Now, as PLS wants nifti volumes in, I can translate vertex coordinates to voxel coordinates and move the CT values into a volume ... which is a bit of a kludge, but will work. Here are my questions: (1) does moving the CT values to a volume sound reasonable, or is there a better way?, and (2) once I have the native CT values in the volume, which PLS analysis option should I use: PET? VBM? I don't believe I want the CT values normalized in case overall thickness is associated with some of my behavioral measures ... if that makes a difference. I would be greatful for any suggestions.
Thanks,
-Jim
Replies:
Untitled Post
Posted on 05/26/09 16:43:57
Number of posts: 394
quote:
Hi All,
I'd like to do a behavioral PLS on native cortical thickness (CT) measures. My CT values are initially stored in one text file for each subject, in which each line of the file has a thickness value (floating point; in millimeters). As our cortical surface is comprised of 81924 vertices, and since thickness is measured at each vertex, we have 81924 CT values per subject.
Now, as PLS wants nifti volumes in, I can translate vertex coordinates to voxel coordinates and move the CT values into a volume ... which is a bit of a kludge, but will work. Here are my questions: (1) does moving the CT values to a volume sound reasonable, or is there a better way?, and (2) once I have the native CT values in the volume, which PLS analysis option should I use: PET? VBM? I don't believe I want the CT values normalized in case overall thickness is associated with some of my behavioral measures ... if that makes a difference. I would be greatful for any suggestions.
Thanks,
-Jim
Hi Jim,Hi All,
I'd like to do a behavioral PLS on native cortical thickness (CT) measures. My CT values are initially stored in one text file for each subject, in which each line of the file has a thickness value (floating point; in millimeters). As our cortical surface is comprised of 81924 vertices, and since thickness is measured at each vertex, we have 81924 CT values per subject.
Now, as PLS wants nifti volumes in, I can translate vertex coordinates to voxel coordinates and move the CT values into a volume ... which is a bit of a kludge, but will work. Here are my questions: (1) does moving the CT values to a volume sound reasonable, or is there a better way?, and (2) once I have the native CT values in the volume, which PLS analysis option should I use: PET? VBM? I don't believe I want the CT values normalized in case overall thickness is associated with some of my behavioral measures ... if that makes a difference. I would be greatful for any suggestions.
Thanks,
-Jim
SInce you already have the data extracted and in vectors for each subject, I would go with the command-line version of PLS. That way you won't have to kludge a conversion to nifti format. You can then dump the output into whatever display utility you wish
Randy
PLS of cortical thickness
I'm Online
Posted on 05/28/09 12:25:42
Number of posts: 8
Hello List,
Thanks for the suggestion Randy. I've never had the courage to try the command line PLS -- and now I have. It all works just fine, however I have a follow-up question.
I am doing a behavioral PLS with 9 behavioral measures and one "condition" (cortical thickness). The bootstrap part of the analysis displays "bootstrap order is not available" ... probably because I only have one condition. Firstly, does this make sense? Secondly, given that the permutation testing found one significant LV, what would be the suggested way of thresholding this LV (in lieu of the bootstrap)?
Thanks,
-Jim
Thanks for the suggestion Randy. I've never had the courage to try the command line PLS -- and now I have. It all works just fine, however I have a follow-up question.
I am doing a behavioral PLS with 9 behavioral measures and one "condition" (cortical thickness). The bootstrap part of the analysis displays "bootstrap order is not available" ... probably because I only have one condition. Firstly, does this make sense? Secondly, given that the permutation testing found one significant LV, what would be the suggested way of thresholding this LV (in lieu of the bootstrap)?
Thanks,
-Jim
Login to reply to this topic.
