Hi Freesurfers,
I managed to project tractography results (seed-based whole-brain tractography performed in FSL) onto the surface of each individual subject (19 OLD, 19 YOUNG).
Each individual surface (including the results) has been transferred to fsaverage to allow statistical group comparison on the group surface level:
mri_surf2surf --srcsubject SUBJ --srcsurfval ${main}/${study}/voxelnum_IN_STRUCT_fs_DILF.mgh --trgsubject fsaverage --trgsurfval ${main}/${study}/voxelnum_IN_STRUCT_fs_DILF_average.mgh --hemi ${side[$i]}h
I concatenated all the subjects (first 19 old, then 19 young) and smoothed it with fwhm of 10 using the following commands:
mri_concat ${main}/${study}/*voxelnum_IN_STRUCT_fs_DILF_average.mgh --o ${main}/${study}/all.mgh
mri_surf2surf --srcsubject fsaverage --srcsurfval ${main}/${study}/all.mgh --fwhm 10 --trgsubject fsaverage --hemi ${side[$i]}h --trgsurfval ${main}/${study}/all_smooth.mgh
My questions:
1) To answer the question where the "tractograophy_results_on_surface" are higher in OLD then in YOUNG I used the follwong command
mri_glmfit --glmdir ${main}/${study}/fs/glmout --y ${main}/${study}/all_smooth.mgh --fsgd ${main}/${study}/fs/glmout/glmout/glm.fsgd --C ${main}/${study}//fs/glmout/omy.mtx --surf fsaverage ${side[$i]}h --cortex
with glm.fsdg
GroupDescriptorFile 1
Class Old
Class Young
Input co_001 Old
Input co_002 Old
Input co_003 Old
Input co_004 Old
Input co_005 Old
Input co_006 Old
Input co_007 Old
Input co_008 Old
Input co_009 Old
Input co_010 Old
Input co_011 Old
Input co_012 Old
Input co_013 Old
Input co_014 Old
Input co_015 Old
Input co_016 Old
Input co_017 Old
Input co_018 Old
Input co_019 Old
Input cy_001 Young
Input cy_002 Young
Input cy_003 Young
Input cy_004 Young
Input cy_005 Young
Input cy_006 Young
Input cy_007 Young
Input cy_008 Young
Input cy_009 Young
Input cy_012 Young
Input cy_016 Young
Input cy_019 Young
Input cy_020 Young
Input cy_021 Young
Input cy_022 Young
Input cy_024 Young
Input cy_014 Young
Input cy_025 Young
Input cy_028 Young
and the contrast file omy.mtx
1 -1
This approach gave me plausible results compared to my pilot FSL analysis.
2) Second I'd like to perform a simple correlation analysis correlating the result_on_surface in OLD with another behavioural measure (e.g. performance in a motor task). Therefore I concatenated the surfaces of all OLD subjects. I wonder how the fsgd file and the contrast file would look like for the GLM_FIT?
If I put
GroupDescriptorFile 1
Class Old
Variables Learning
Input co_001 Old x
Input co_002 Old x
Input co_003 Old x
Input co_004 Old x
Input co_005 Old x
Input co_006 Old x
Input co_007 Old x
Input co_008 Old x
Input co_009 Old x
Input co_010 Old x
Input co_011 Old x
Input co_012 Old x
Input co_013 Old x
Input co_014 Old x
Input co_015 Old x
Input co_016 Old x
Input co_017 Old x
Input co_018 Old x
Input co_019 Old x
in the fsdg with x = behavioural result
and " 0 1 " in the contrast file corr.mtx (looking for positive correlation between surface_result and behaviour), I don't get any significant results. And I am not sure whether I performed the analysis correctly.
Can anybody tell me whether the GLM, fsgd and contrast files are set up correctly, both for the group comparison in 1) and for the correlation analysis in 2).
Thank you very much!
Kind regards, Robert
--
Pflichtangaben gemäß Gesetz über elektronische Handelsregister und Genossenschaftsregister sowie das Unternehmensregister (EHUG):
Universitätsklinikum Hamburg-Eppendorf; Körperschaft des öffentlichen Rechts; Gerichtsstand: Hamburg
Vorstandsmitglieder: Prof. Dr. Martin Zeitz (Vorsitzender), Dr. Alexander Kirstein, Joachim Prölß, Prof. Dr. Dr. Uwe Koch-Gromus