> I know that this is a subject that has come up once or twice, but I
> can't seem to find a direct answer to my query in the archives.
>
> The query is with regard to normalisation in the optimsed method. I am
> performing analyses on 50 patient and (versus) 50 control scans. Do I:
> (1) normalise the patient GM images to a mean GM image constructed
> from the normlised, segmented and smoothed patient scans from the
> standard protocol, whilst normalising control GM img files to a mean
> control image constructed in the same way, or
> (2) normalise all individual patients and controls to a mean GM image
> constructed from all patients and controls?
The latter. Normalise all images to the same template in order to reduce
the effects of systematic differences between the controls and
patients.
>
> How would (1) affect the comparisons?
Normalising one set of data to one template and the other set to another
would result in some of the systematic differences between the templates
appearing in your final results. The latter approach should produce the
most internally consistent registrations.
> What effects would (2) have on the results?
There may be slightly more within group variance. The controls are likely
to be less well matched with each other, and the patients may be less well
matched with each other.
>
> I hope that this was clear!
I hope my answers made sense.
Best regards,
-John
--
Dr John Ashburner.
Functional Imaging Lab., 12 Queen Square, London WC1N 3BG, UK.
tel: +44 (0)20 78337491 or +44 (0)20 78373611 x4381
fax: +44 (0)20 78131420 http://www.fil.ion.ucl.ac.uk/~john
|