Hi,
I'm trying to analyse ReHo with a one sample t-test and I'm not sure if I'm doing it correctly :-/. At first I generated a group mask with 90 %. Then I used the statistical analysis toolbox for calculating a one sample t-test with the swz ReHo-maps and used base 0 as I read in this post: http://rfmri.org/content/one-sample-t-test-results-reho.
Next, I used the dpabi-viewer to overlay the t-map on the ICBM map (see picture attached), but now I'm stuck. I was trying to do a FDR/GRF/Alphasim respectively, but I don't know what mask-file to use here. Do I need to use the group mask for this or some other file?
Also, it seems to me that my coverage is not that good (see picture) so I will have trouble identifying which brainarea is active. What can be done in such a case?Using a 50 % group mask provides better coverage, but I wonder if there are some other possibilities :)
One last question: When calculating a two sample t-test, e.g. for comparing a control and a patient group. Do I need to generate one group mask for controls and patients together or don't use any mask at all?
I appreciate the help & thanks in advance. Btw the dpabi viewer is really nice :)
David
Attachment | Size |
---|---|
rehomaps.png | 165.5 KB |
Hi David,
Hi David,
You should give your group mask.
I think the lowest percentage is 80%. If a voxel doesn't show in 80% of subjects, the statistical power will be an issue.
Best,
Chao-Gan
..
Thank you! Is it possible to generate a group mask only for grey matter voxels?
That's because after calculating a two-sample t-test to compare the amygdala fc-map or reho-maps asf. between patients and controls I get significant voxel which somehow (or some of them) do not lie within grey matter and I would like to limit the analysis to grey matter voxels only.
peace
David
Hi David,
Hi David,
To generate a group specific GM mask, I will first average all the subjects' GM maps (wc1*) and then give a threshould (e.g., 0.2).
If you want to make sure the GM voxels have EPI signals, then you can use the abovementioned mask to multiply the group mask you have generated already by DPABI.
Best,
Chao-Gan
...
I see, thank you very much :)
masking toolbox
for all who are interested. I found this very helpful spm plugin for creating a mask and finding an optimal threshold:
http://www0.cs.ucl.ac.uk/staff/g.ridgway/masking/