Hi
On 04.05.2011, at 17:08, Kathryn Cullen <[log in to unmask]> wrote:
> Hello, I have conducted dual regression on an fMRI dataset after having done an ICA with MELODIC. I have 2 questions:
>
> 1. Some of the stage 3 tcfe_corrp_tstat results have a ton of voxels. What is the unit of the voxels? It seems to range from 0 to 1. I was trying to fiddle to see if I could threshold somehow but didn't know the scale
These are corrected p values (well 1-p values to be precise, see the online documentation for randomise)
> 2. I have read the Filippini 2009 paper "Distinct patterns of brain activity in young carriers of the APOE-4 allele" in which a local FDR correction was conducted after randomise. Another paper that has done this was Veer et al 2010, "Whole brain resting-state analysis reveals decreased functional connectivity in major depression". I'm not sure if this is appropriate to do in every case, what do you think?
Clearly not, it will depend on whether you think cluster correction makes sense given your particular investigation.
> SHould all dual regression group atscomparison randomise analyses be followed up by additional correction with FDR?
Note that in both cases the local FDR was run on tstats, not on corrected maps
> If this is recommended, how can I do it? I'm not clear from the papers which mask to use for the fdr.
The local fdr is not available right now.
Hth
Christian
> THanks very much for your help,
>
> Katie Cullen
> University of Minnesota
|