Dear Nicola,
I always find the ample illustrations in the following paper very useful to form an idea of the likely effect size of any age differences:
R.S.N. Liu, L. Lemieux, G.S. Bell, S.M. Sisodiya, S.D. Shorvon, J.W.A.S. Sander, and J.S. Duncan
A longitudinal study of brain morphometrics using quantitative
magnetic resonance imaging and difference image analysis
NeuroImage 20 (2003) 22-33
The main effect seems to be perceptional - whether you ask someone who's 25 or 55 will make the most major difference where they perceive an effect of brain volume loss to be starting :-).
On the whole and for the sake of consistency between centres (as you say), I'd personally probably go for option 1). But then I'm not an fMRI person, and I don't know whether you're chasing an effect that really does show very pronounced changes with very small age increments.
HTH, A
-----Original Message-----
From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On Behalf Of <Nicola> <Canessa>
Sent: 16 October 2007 09:06
To: [log in to unmask]
Subject: [SPM] normalisation of 40/50-years-old subjects' brains
Dear experts,
we are using SPM5 to compare patients vs. controls with fMRI. Our patients
do not have gross anatomical abnormalities (this is one of the exclusion
criteria). Yet, since the mean-age of both patients and controls is about 43
± 8 (a little bit elder than the volunteers from which the MNI template was
created) we are wondering whether we should use a specific template for
normalization, and we are trying to distinguish among 3 options:
1) to run the classical pre-processing and simply taking the deformation
parameters for normalization from the segmentation of the individual T1
images (coregistered to the mean of the realigned functional images). In
this case, one further option would be to use the "no priors" option of the
VBM-toolbox 5.1
2) to search for an age-specific template. Provided that such templates do
exist/are available somewhere, probably this would be the option for the
best normalization as possible, yet the results would not be in the MNI
space, and thus they could not be directly compared with those from other
studies. Therefore, I would be inclined to discard this option.
3) to create an ad-hoc-template, but since we don't have hundreds of images
(just 15 patients and 15 controls) this does not sound feasible
Any suggestions/criticism about these options?
Many thanks in advance,
best wishes,
nicola
***************************
Nicola Canessa, Ph.D.
CRESA, www.cresa.eu
San Raffaele Research Institute,
Via Olgettina 58,
20132, Milano, Italy
+39-02-2643-4888
[log in to unmask]
***************************
|