Hi Mustafa,
I assume that you used the data-demeaning option in randomise as the
model looks like it's demeaned.
Your data correlates globally with the EV - the histogram shows a
large positive shift. Looking at the maxc image (corrected cluster p-
values) you do indeed have a significant area in the posterior of the
corpus callosum - I'm not sure why you think you didn't get anything?
Also, almost certainly if you reduce the cluster-forming threshold
(to say 2.5 or even 2) you'll probably find the cluster enlarged and
still significant.
Wrt correcting for age - in general in such analyses this is not safe
as you can't necessarily assume that FA (etc) correlates linearly
with age. However, as long as the age covariate is reasonably
orthogonal to the EV of interest then it's hopefully ok - needs
careful thinking though.
Cheers, Steve.
On 30 Apr 2007, at 14:16, Mustafa Anjari wrote:
> Dear Prof Smith
>
> We have some DTI data that we’ve processed for TBSS analysis. We have
> generated design matrix and contrast files for the subsequent
> voxelwise
> analysis of the data using GLM, in order to perform linear
> regression on the
> data based on a clinical variable. When we then analyse the data with
> Randomise (using n = 5000, c = 3), the unthresholded t-stat image
> shows
> regions of reduced FA that correlate with reduced gestational age
> (t > 3, p
> < 0.05). Looking at the multiple comparison-corrected cluster p-
> value image,
> however, this difference does not appear significant in any brain
> regions (t
>> 3). Do you think we are being too harsh in our multiple comparison
> corrections? Would it be possible for you to have a look at our
> data? I’ve
> uploaded the t-stat images and the design matrix and design
> contrast files
> (upload ID 680318).
>
> Also, are we correct in thinking that we can correct for age at
> scan by
> including this as a second EV and demeaning the data?
>
> Thank you very much for your help.
>
> With best wishes,
> Mustafa
>
> ----------------------------------------------
> Mustafa Anjari
> Robert Steiner MRI Unit
> Imaging Sciences Department
> Imperial College London
> Hammersmith Hospital
> Du Cane Road
> London W12 0HS
> UK
>
> Tel: +44 20 838 32336
------------------------------------------------------------------------
---
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director, Oxford University FMRIB Centre
FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
------------------------------------------------------------------------
---
|