Dear Experts
I have few questions:
I'd like to use AlphaSim to obtain the minimum cluster-size (for multiple corrections) for my ROI analysis of ppi, and done the following:
I wonder if these are correct.. Could you clarify my mistakes?
1) FWHM
I did the below at “smoothness Estimation” window that popped up by pressing “Estimate” in the “REST AlphaSim” window.
I selected SPM T.img file at “Input statistical Map” which was obtained during the 2nd level analysis in spm8 (for the ppi)
And then, I selected “dlpfc.nii” file at “Input Mask File” (This ROI mask was created by PickAtlas toolbox implemented in spm8)
By clicking the “Run” bottom, I got the values of: FWHMx=7.68.., FWHMy=8.75.., FWHMz=11.48.
As a side note, the smooth Kernel that I used during the preprocessing in spm was 8mm
---------------------
# Subsequently, I did the following at the “REST AlphaSim” window (this is after pressing the “Run” bottom stated above).
I wonder if these are also correct;
2) rmm
I entered the value 3, according to the equation of: 1.412 L < rmm < 1.732 L that I obtained from the source below
http://afni.nimh.nih.gov/afni/doc/faq/24
In this equation, I set the L as "2", since our voxel size is 2*2*2.
Please note that my original voxel size during the scanning was 3.5*3.5*3.5, but this was resampled to 2*2*2 by Normalization during preprocessing in spm.
3) Mask
I selected “dlpfc.nii” at “Mask” which is identical to the ROI mask stated above.
4) clustersize
By pressing “Run” after I entered the values of 0.05 and 1000 for “P threshold” and “iterations”, respectively, I got the AlphaSimtext.txt.
Within this file, in a “alpha” column where it states 0.01, the corresponding row of the “CI size” column states the value 189.
I assume that this is the number of minimum voxels that I am looking for p < 0.01
-----------------------
(I used the Rest toolbox ver1.8 20130615, and my matlab is 2014a)
Thank you so much.
Best
Shisei
Hi Shisei,
Hi Shisei,
I didn't see anything wrong with the procedures you have described.
Usually you can set p < 0.05 at cluster level (i.e., after correction).
Another note is that beyond estimating smoothness from the statistical image, you can also estimate from the 4D residuals during statistical analysis.
Best,
Chao-Gan
RE
Dear Chao-Gan
Thank you so much for your prompt reply!
I appreciate you helpful comments.
I would like to study them carefully.
Sincerely,
shisei