[MINC-users] ROI stats

[Andrew Janke]

On Tue, Feb 19, 2008 at 2:27 AM, John Absher <john at absherneurology.com> wrote:
>  4 subjects, each had baseline CBF PET scans x 2, then they had
>  low-blood-pressure CBF PET scans x 2.
>
>  I have run the SPM2 analysis, but I also wanted to look at the ROIs.  I have
>  data for 115 ROIS, for each of the 4 PET scans obtained for each of the 4
>  subjects.

Are these 115 ROI's the same for each subject?  ie: did you draw them
or are they obtained by some automatic method?

>  What is the best statistic and correction method to use for evaluating the
>  significance of the regional changes in CBF when baseline measures are
>  compared to low-BP measures?

Well if understand you would have 115 mean values of CBF for each
"experiment". You would then I presume average the results of each of
the baseline and low-blood-flow images at which stage you would have
115 difference measures?

You would then have 115 difference across 4 subjects so at this stage
I presume you just need to discriminate which of them is more/most
predictive of the baseline vs low-blood-flow condition.

Do I have it right? :)  I guess the question of which stats would
depend on what you want out of this.  Do you want to know which ROI is
the best or where in each ROI is the best?  The former means a simple
stats analysis and the latter would get a tad more complicated (and i
suspect underpowered with only 4 subjects).


-- 
Andrew Janke (a.janke at gmail.com || http://a.janke.googlepages.com/)
Canberra->Australia +61 (402) 700 883