Ce,
You can use EasyROI that will extract your betas for sphere/box/cluster and save .mat with all the values (ie in each voxels, thus you have your voxel count as well) as well as mean and 1st eigen value.
For percent signal change, beta*100/last beta of the design matrix*max design matrix (because it's not 1 after convolution) .. see http://www.sbirc.ed.ac.uk/cyril/html/BOLD_percentage.html
Good luck
Cyril
Dr Cyril Pernet
BRIC / SINAPSE
University of Edinburgh
----- Reply message -----
From: "Ce Mo" <[log in to unmask]>
Date: Tue, May 22, 2012 09:47
Subject: [SPM] voxel sizes and ROI
To: <[log in to unmask]>
Dear all,
I have 2 questions regarding voxel sizes and ROI.
Suppose I have a spheric ROI with a radius of 10 mm centered around a MNI coordinate, say [x1,y1,z1].
I am interested in how many voxels there are in this ROI and above all, their individual values.
The intuitive thought is to find all these voxels based on the distances between the sphere center and the candidate voxels the MNI coordinates. However, I am not so sure whether I should take voxel sizes into consideration, and more importantly, whether this approach is feasible.
Or is there anybody who has better ideas?
Moreover, I understand that softwares such as Marsbar can be used to extract the mean parameter (beta) value, I suppose that will be to sum all the values of the voxels within that ROI and then divide it by the number of voxels to obtain the mean value. However, I wonder whether there are any additional scaling so that the extracted mean can be interpreted as percent signal change.
Many thanks and best regards,
Ce
