Dear FS experts,
I are analysing longitudinal data - the difference of the cortical thickness, surface area and volume between 2 groups (1 and 0) with 2 time points for each group (0, 1 --all the same time interval: a month).
I followed the page http://surfer.nmr.mgh.harvard.edu/fswiki/LinearMixedEffectsModels in wiki. According to experts' advice in the mail-list, I used voxel-wise mixed model “lhstats = lme_mass_fit_vw(X,[1],Y,ni,lhcortex)” to do analysis. And I got the result “sig.mgh” (Freesurfer significance map -log10(p-value)).
1. I loaded the “sig.mgh” to do the FDR correction in the qdec tool. When I press the button of ‘set using FDR ’, the threshold value changes. But the minimum value in the color bar is a little bit small than the value in the table. I used the value in the color bar to get the correction result.
Q: Can I do FDR correction like this? And could you give me some more advices about the correction? And how to do other multiple comparison correction like Monte Carlo simulation in longitudinal analysis?
2.To get more infotmation about the cluster, I used the cmd “mri_surfcluster --in sig.mgh --subject fsaverage --surf inflated --hemi lh –thmin *( minimum value in above FDR correction) --sign abs –annot aparc –-sum lh.t.cluster.summary” to generate the set of detected clusters and their anatomical coordinates.
Q: I want to know if I use the correct command?
3.I saved clusters’ label file manually in qdec tool. Then I use the cmd "mri_label2label" and "mris_anatomical_stats" to get the value of thickness. Just like the page: http://surfer.nmr.mgh.harvard.edu/fswiki/FsTutorial/AnatomicalROI
For area and volume, can I get in the same way? For example, in the stats:# ColHeaders StructName NumVert SurfArea GrayVol ThickAvg ThickStd MeanCurv GausCurv FoldInd CurvInd
lh.1.label 25 19 51 3.044 0.237 0.105 0.068 0 0
I think that the “SurfArea” is the value of area, “GrayVol” is the value of volume. I wonder that it may be wrong way. In my other GLM analysis, I found that cortical surface area or volume is not an integer in the file “…y.ocn.dat”.
Q:Is it right way to get the cluster value? What methods do you recommend to extract the cluster’s mask and the value within the cluster?
Thank you very much, and looking forward to reply.
Kind regards,
Livia
On 6/19/17 10:08 PM, Livia Liu wrote:
Dear FS experts,
I are analysing longitudinal data - the difference of the cortical thickness, surface area and volume between 2 groups (1 and 0) with 2 time points for each group (0, 1 --all the same time interval: a month).
I followed the page http://surfer.nmr.mgh.harvard.edu/fswiki/LinearMixedEffectsModels in wiki. According to experts' advice in the mail-list, I used voxel-wise mixed model “lhstats = lme_mass_fit_vw(X,[1],Y,ni,lhcortex)” to do analysis. And I got the result “sig.mgh” (Freesurfer significance map -log10(p-value)).
- I loaded the “sig.mgh” to do the FDR correction in the qdec tool.
When I press the button of ‘set using FDR ’, the threshold value changes. But the minimum value in the color bar is a little bit small than the value in the table. I used the value in the color bar to get the correction result.
Q: Can I do FDR correction like this? And could you give me some more advices about the correction? And how to do other multiple comparison correction like Monte Carlo simulation in longitudinal analysis?
QDEC is slowly becoming non-functional, and I think the FDR interface may be faulty (or just hard to use). In version 6, I created a program called mri_fdr that you can use to compute (and/or apply) the threshold.
2.To get more infotmation about the cluster, I used the cmd “mri_surfcluster --in sig.mgh --subject fsaverage --surf inflated --hemi lh –thmin *( minimum value in above FDR correction) --sign abs –annot aparc –-sum lh.t.cluster.summary” to generate the set of detected clusters and their anatomical coordinates.
Q: I want to know if I use the correct command?
Don't use the inflated surface as that does not correspond to a physical surface (the default is white, which is probably what you want). Otherwise it looks ok. Note that you can pass it the FDR with --fdr, and it will compute the threshold for you. Also, I would include the mask with --mask to mask out the non-cortical areas.
3.I saved clusters’ label file manually in qdec tool. Then I use the cmd "mri_label2label" and "mris_anatomical_stats" to get the value of thickness. Just like the page: http://surfer.nmr.mgh.harvard.edu/fswiki/FsTutorial/AnatomicalROI
For area and volume, can I get in the same way? For example, in the stats:# ColHeaders StructName NumVert SurfArea GrayVol ThickAvg ThickStd MeanCurv GausCurv FoldInd CurvInd
lh.1.label 25 19 51 3.044 0.237 0.105 0.068 0 0
I think that the “SurfArea” is the value of area, “GrayVol” is the value of volume. I wonder that it may be wrong way. In my other GLM analysis, I found that cortical surface area or volume is not an integer in the file “…y.ocn.dat”.
Q:Is it right way to get the cluster value? What methods do you recommend to extract the cluster’s mask and the value within the cluster?
what is your command line?
Thank you very much, and looking forward to reply.
Kind regards,
Livia
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Douglas Greve -
Livia Liu