Hi,
We are doing a group analysis of ADC maps coming from DWI images.
(two-sample t-test, stroke patients vs controls)
We guessed that we should use a global normalisation since the ADC
global value is different for each subject.
However, there are two options under 'global normalisation" and we don't
know wether we should use "proportional scaling" or ANCOVA.
We have tried the three possible options and found the following results:
- using "proportional scaling", we found bigger clusters than with "no"
but with same peaks
- using "ANCOVA", we found even bigger clusters, with the same first
peak but with different other peaks
The only messages that we found on the SPM list related to global
normalisation concerned VBM or PET data analysis.
Does anybody have an experience for the specific cases of ADC analysis ?
Any help would be greatly appreciated !
Many thanks,
Olivier
|