Dear Kim,
Usually I would go with AnCova and only use proportional scaling when
it is know that there is a (little) fluctuation in injected dose, but
this answer is from a water-PET perspective.
However, in your case you might be limited in your degrees of freedom
as you have only one or two scans per subject and you already include
covariates, so you might be bound to using proportional scaling. If
you still have enough power I do not see any harm in using AnCova by
subject assuming that you have controlled the injected dose.
Kind regards,
Simone.
On Sat, May 3, 2008 at 5:40 PM, ±θΑφΗφ <[log in to unmask]> wrote:
> Dear experts
>
> Apologize for the basic question.
> I'm analyzing FDG-PET scans in 25 patients with a neurological disorder.
> In addition to group comparison (patients vs controls), I want to correlate
> regional glucose metabolism with clinical parameter such as disease
> duration.
> I chose 'single subject_covariate only' module in PET models (spm2), with
> disease duration as covariate-of-interest and age/sex as nuisance variables.
> In the next step of 'Select global normalization', there are two options,
> Ancova vs. proportional scaling. Which one is appropriate for my study?
> I have tried each method but the results are very different.
>
> Accodring to an spm99 manual in Website, there is the following sentence,
> "AnCova is advised for multi-subject studies unless differences in global
> flow are large (e.g., due to variability in injected tracer dose). Because
> AnCova also uses one degree of freedom for each subject/group, proportional
> scaling may be preferable for single-subject studies."
> Does the selection of 'Ancova' is more suitable for my design?
> Many thanks for the good advice.
>
> Kim
--
Dr. A.A.T. Simone Reinders, MSc PhD
King's College London
Institute of Psychiatry (IoP)
Box P063, De Crespigny Park
London SE5 8AF
United Kingdom
|