Dear SPM users,
Many papers, including those in prominent journals, report fMRI results small-volume corrected (SVC) but present figures at an uncorrected threshold for display.
However, I am interested in how results can be displayed "corrected", if small-volume correction (SVC) is used within a region of interest.
(I apologize for the possibly redundant question, but I haven't found a clear answer to this question on the list or among peers.)
1. One option that seems to be suggested by a prior posting is promising:
In SPM, from the cluster-corrected SVC results table on the bottom-left, one could use the number of voxels listed by FWEc and display results uncorrected at the initial whole-brain threshold that are in clusters that larger or bigger.
2. Alternatively, but probably incorrectly:
One could take the number of voxels in a significant cluster listed in the SVC results table and derive an uncorrected threshold which would leave a cluster of that size (e.g. p may be <.0026 may instead of <.005). The reasoning here is that those # of voxels went into determining significance, so that's what should be displayed.
These approaches may be limited to the particular cluster in question or have other issues that I'd really appreciate hearing comments about. I know that for both, the cluster size depends on the initial whole-brain uncorrected threshold (p<.001/p<.005/etc). And neither work if one is using peak-level correction (for small but strong clusters, sometimes FWEc = Inf), so perhaps there is a different method that would work here, too.
Does anyone have any solutions, guidelines, or comments?
Thank you so much!
Best regards,
Elliott
Learning Lab
Columbia University
312 Schermerhorn Hall
[log in to unmask]
|