Hi Michal,
My understanding of small volume correction (see e.g.
http://imaging.mrc-cbu.cam.ac.uk/imaging/SmallVolumeCorrection) is
that it only relates to the estimated resel count as used for Gaussian
Random Field theory FWE correction.
FDR as I understand it (from e.g.
http://www.sph.umich.edu/~nichols/FDR/) does not make use of the resel
count, so I don't think SVC is relevant.
A quick glance at the code seems to support this, since spm_uc_RF (the
random field FWE correction) needs to be passed the resel count, while
spm_uc_FDR does not include an argument for resels.
I hope that's correct, and helpful. As for why your FDR corrected p
values are fixed and high, I'm afraid I can't think of anything to
suggest.
Best,
Ged.
Michal Assaf wrote:
> Hello all,
>
> We would like to do a small volume correction in SnPM, using a mask
> image. Since the SVC toolbox is not available for SnPM, we masked our
> individual contrast images with the mask and then entered them into a
> group analysis in SnPM. However, for some reason this modification
> leaves us with a fixed (and high) P-FDR values for all clusters (while
> the uncorrected and FWE values vary between clusters). Since we wanted
> to use the FDR correction, I was wondering if anyone has implemented a
> small volume correction for FDR in SnPM?
>
> Thanks in advance for your help!
> Michal
>
> Michal Assaf, M.D.
> Senior Research Scientist
> Olin Neuropsychiatry Research Center, Institute of Living
> Assistant Professor Adjunct
> Department of Psychiatry, Yale School of Medicine
> 200 Retreat Ave., Hartford, CT 06106
> Tel: 860.545.7792 | Fax: 860.545.7797
>
>
>
>
>
|