Dear Vinci,
for a single voxel you can calculate r using the following formula:
r = sign(t)/(sqrt(df/(t*t)+1))
To transform a spmT image to a correlation map you can use the attached tool.
Best regards,
Christian
____________________________________________________________________________
Christian Gaser, Ph.D.
Assistant Professor of Computational Neuroscience
Department of Psychiatry
Friedrich-Schiller-University of Jena
Philosophenweg 3, D-07743 Jena, Germany
Tel: ++49-3641-935805 Fax: ++49-3641-935280
e-mail: [log in to unmask]
http://dbm.neuro.uni-jena.de
On Wed, 17 May 2006 09:27:42 +0100, Vinci Cheung <[log in to unmask]> wrote:
>Dear SPM friends,
>
>1) I wonder how to get the r from the SPM output of the Simple Regression
>when I have put IQ as covariate, and got a few cluster with sigificant +ve
>correlation with DTI FA values. The voxel level uncorrected p is .0001,
>FDR-corrected p is .09, cluster level corrected p is .03. I search the
>archives and know that T cannot be used to transform back to r.
>
>2) I use MARBar to extract the mean of ROI from these significantly +ve
>correlated clusters with IQ. I wonder if the mean extracted are mean FAs
>or mean r of a significant cluster. If what I got is mean FAs, can I put
>this to SPSS and correlate these mean FAs with IQ to get back the
>cluster's correlation r with IQ??? And report this Pearson r as a
>cluster's r with IQ??
>
>3) Regarding controlling for multiple comparison, is the below threshold
>good enough: voxel level uncorrected p<.001 & voxel size >150, or cluster
>level corrected p<.05?
>
>4) I read from previous messages that the covariates need to be mean
>corrected. But I wonder if I need to do that for my IQ to look at +ve & -
>ve correlation with normalized DTI FA maps (coregistered B0 to EPI
>template)?
>
>5) Any study to show the use of Simple regression in SPM2 or method paper
>to justify the use of simple regression to look at correlations with brain
>data that I can quote?
>
>
>
>Thanks!
>Vinci
>Department of Psychiatry
>The University of Hong Kong
>Hong Kong
>===========================================================
==============
|