Dear Yann,
On Wed, 31 Aug 2016 12:21:00 +1000, Yann Quidé <[log in to unmask]> wrote:
>Dear Christian and other CAT12 users,
>
>First, once again, thanks for the great tools. However, I have few questions:
>
>1) I encounter an error when trying to extract additional surface parameters (ie gyrification index etc.). The error message says:
>
>SPM12: cat_surf_parameters (v964) 08:09:26 - 31/08/2016
>========================================================================
>Extract parameters for <mypath>/surf/lh.central.myname.gii
> Display <mypath>/surf/lh.gyrification.myname
>{Reference to non-existent field 'rdata'.
>
>Error in cat_surf_parameters (line 74)
> roijob.rdata = job.rdata;
>}
>
>Did I miss something obvious?
No, this is my fault. There was a developer option wrongly placed in that tool (if you call it in developer mode it will not crash). This issue should be solved in the new release r991. Please update to that version.
>
>
>2) When extracting cortical ROI thickness, output for the a2009s atlas gives me Lunknown and Runknown as the 2 columns following my subject ID. Similarly, within the DKT40JT, L/Runknown are present and I have NaN both the L/R bank of STS and L/R corpus callosum. Are Nan values normal for these regions with CAT12? What do the L/R unknown regions stand for?
Unknown areas are all areas that are not defined in those atlases. Please check the papers about the atlases for more information. I have simply used the original freesurfer atlases:
https://surfer.nmr.mgh.harvard.edu/fswiki/CorticalParcellation
The NaN-values are below the corpus callosum and around the ventricle areas. These areas are mot meaningful to analyze and are set to NaN in order to not calculate any statistic in those regions. Thus it might be that these regions are defined in the atlas, but not used in the analysis.
>
>3) When looking at the vROIs, what atlas would you recommend to use (especially for accuracy in subcortical regions)?
Personally, I would prefer the DKT40 atlas. The Destrieux atlas is using too many small regions that are some times not very well divided if you check the overlay onto the freesurfer average brain.
>
>
>4) Just for confirmation, I was looking at results from a multiple regression analysis on cortical thickness. I followed the manual for Basic Models on my s15mm files (ie did not use TIV or threshold masking), then used the CAT12 Estimate module, and finally used the SPM12 contrast manager and visualised my results in SPM12 (looked at both p=0.001 uncorr + cluster FWE correction and voxelwise pFWE=0.05). Is this correct?
>What surprises me is the amount of “positive” results! Did I miss a step?
Difficult to say if I can't see the results and don't know something about your analysis and sample. Your steps sound fine for me...
Best,
Christian
>
>
>Thanks for your time.
>
>Cheers,
>
>Yann
|