Dear Andrea,
| In a PET-study if I want to compare an amnesia-patient at rest with a
| normal subject at rest, have I to make sure that it has been used 18-FDG
| or O-15 in both sides? Are there other shrewdnesses to conside?
In such a comparison you are essentially estimating the distribution of
normals, and then looking at that distribution to see if the patient
data could reasonably have come from it. (This is usually taught as a
"prediction interval", but turns out to be the same as a two-sample
t-test with only one member in the second group, which is what SPM
implements.)
Thus it is essential that if the patient were in fact normal, they
would give a PET image as would be expected of a Normal. Thus, the
patient must be examined in exactly the same way as the normal
controls. Clearly different tracers give different distributions of
intensity, so you *must* use the same tracer. Further, there should be
no systematic differences in scanning protocol or subject treatment
between the normals and the patient. (This includes different placement
within the tomograph, pharmaceutical influences, instructions etc...)
Failure to match the scanning conditions may result in observed
differences that are purely an artefact of the different situation.
In addition, the better you control the variability of non-systematic
differences between subjects (i.e. head counts, position in scanner),
the more powerful the comparison of patient vs. normal controls will
be.
Lastly, you must use a random effects analysis to assess these data: A
simple way to do this is to average any repeated measurements within
subject to give a single average rest image per subject. With one
(average) scan per subject, a standard SPM two sample t-test analysis
is in fact a repeated measures random t-test accounting for random
subject effects.
Hope this helps,
-andrew
+- Dr Andrew Holmes [log in to unmask]
| -___ __ __ Wellcome Department of Cognitive Neurology - |
| ( _)( )( ) Functional Imaging Laboratory, Stats & |
| ) _) )( )(__ 12 Queen Square, Systems |
| (_) (__)(____) London. WC1N 3BG. England, UK |
+---------------------------------------http://www.fil.ion.ucl.ac.uk/-+
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|