Hi Andrew (and SPMers),
> Can anyone help me figure out which is the correct resel size to use
> when calculating the corrected p-value for a cluster of active voxels
> in a second level SPM2 analysis? Our first level analyses output 10
> contrast images (one for each subject) which serve as the input for
> our second level analysis. By default SPM calculates the corrected p
> in the second level analysis using the resel size from the 10
> contrast images, but my colleagues and I were wondering whether it
> would make more sense to use the resel size from the original
> filtered BOLD images that serve as the input to the first level
> analyses. Thoughts?
OK, here are my thoughts. You should use the resel size from the
2nd-level analysis. The resel size should reflect the smoothness of the
images you are analyzing. T-images or F-images in your 2nd-level
analysis are calculated based on 10 contrast images, thus the resel size
should reflect the smoothness of these 10 contrast images.
> Given that we are wanting to use cluster-level corrected p-values and
> have 10 subjects, would folks advise us to use SnPM for our second
> level analysis and the more sophisticated cluster-level analyses
there?
Well, you can do an SPM analysis with 10 subjects and the analysis is
perfectly valid. However, SnPM is known to have better sensitivity than
SPM when df is small. You might want to consider using it.
Satoru Hayasaka PhD ----------
Assistant Professor, Public Health Sciences & Radiology
Wake Forest University School of Medicine
(ph) +1-336-716-8504 / (fax) +1-336-716-0798
(email) shayasak _at_ wfubmc _dot_ edu
|