Hello,
We have a question regarding 3dClustSim and the mask used for processing.
In the past we would make a group mask for group analysis. The analysis was conducted in tlrc space.
But I recall in the past it was recommended to resample the group-mask voxel size to the orignal space voxel size before running AlphaSim:
"As data are acquired in original voxel size, do make sure that Monte Carlo simulations are done in original voxel size (but not original brain shape - you will see the difference later) instead of higher resolution such as tlrc space. This is to make sure cluster formation is properly simulated"
[
afni.nimh.nih.gov]
Does this still apply using 3dClustSim.
Here is the following output using a group mask in standard space vs the resample group mask in original voxel size
Orig voxel size:
# 2-sided thresholding
# Grid: 52x63x53 3.03x3.03x3.00 mm^3 (60011 voxels in mask)
#
# CLUSTER SIZE THRESHOLD(pthr,alpha) in Voxels
# -NN 1 | alpha = Prob(Cluster >= given size)
# pthr | .10000 .05000 .02000 .01000
# ------ | ------ ------ ------ ------
0.050000 8.1 8.7 9.5 10.1
0.025000 5.3 5.7 6.2 6.6
0.010000 3.7 3.9 4.3 4.5
Tlrc voxel size:
# 2-sided thresholding
# Grid: 79x95x79 2.00x2.00x2.00 mm^3 (206345 voxels in mask)
#
# CLUSTER SIZE THRESHOLD(pthr,alpha) in Voxels
# -NN 1 | alpha = Prob(Cluster >= given size)
# pthr | .10000 .05000 .02000 .01000
# ------ | ------ ------ ------ ------
0.050000 9.2 9.8 10.6 11.2
0.025000 6.0 6.4 6.9 7.3
0.010000 4.1 4.4 4.8 5.0
Which output is the correct output to use?
If the orig size voxel method is correct, what is the purpose of resampling to original size voxel? How does minimum cluster size criteria translate if the analysis is done with tlrc size voxels?
Any help would greatly be appreciated.
Thank you for your time.
Michael
