>Ananth
>
>I'm not sure to understand your questions - I guess there are  
>different issues here ..
>
>> I am trying to discern whether there is difference in the amount of
>> activation between two drug conditions. The conditions are two different
>> drugs to the same autism patient population so that the I need to figure out
>> if there is any difference between autism patient population drug 1 and
>> autism patient population drug 2. How would I go about doing that?
>
>so you have 2 groups of subjects trying 2 different drugs - do you  
>have some pre- post- treatment scans? otherwise I can't see how to  
>analyze the data since drug and group effects are confounded ..  
>assuming pre- post- I would set up an anova using the flexible  
>factorial with factor subjects, pre/post and drug/group and then model  
>subjects and the interaction - on the resulting 4 columns pre-gpA  
>pre-gpB post-gpA and post-gpB a F contrast -1 1 1 -1 will tell you if  
>post-gpA vs post-gpB are different given the possible diff in pre  
>treatment.

I think I wasn't very descriptive with the study design. The scans are all
post treatment. No pre-treatment drugs. There are four drugs, D1 to D4. The
subjects are administered each drug, and are scanned after waiting for the
wait period prescribed for each drug. At present, I am only investigating
two drugs at a time because my understanding is not sophisticated enough to
do all four at a time. So the following question is for comparing between
say D1 and D2. 

I know I can find the group activation maps using the 2nd level spec >
estimate and the beta1 (block design task - rest-activation-rest model)
images from the 1st level spec > estimates. I am unclear on what the
statistics (the numbers that show up below the glass brain image), once I
have pushed the "whole brain", "current cluster" or "small volume" buttons
mean. 

Suppose I have 10 subjects, what are those statistics (the same numbers that
show up below the glass brain image) going to say about each individual
person's scan? I tried searching for this, as I am sure it was answered, but
there is so much info that I almost gave up. I would greatly appreciate a
reply. I am using SPM8. 

>> Thus far,
>> I have been using "small volume" button centered around the ROI for each
>> drug condition.
>
>using small volume is just a way to control the search space and, to  
>my opinion, is valid only then you have prior hypothesis where to  
>search (or performed an orthogonal contrast to define where to search)  
>- i don't think this is the solution you are looking for ..
>
>> The problem I've encountered is, when I click that button with the
>> appropriate ROI, say (-42, -62, -16) for left fusiform, and say sphere, I
>> don't know what each one of the statistics are telling me. There is a set
>> level, cluster level and peak level statistics. What do those mean? I assume
>> that k sub E in cluster level is actually the number of voxels. This number
>> varies depending on whether I select FWE or not in the results section. What
>> is more appropriate for my study? Is it sufficient to use k sub E in the
>> cluster-level at the individual statistic to compare between different drug
>> conditions? The manuals are not helping me at this. I would greatly
>> appreciate an explanation on this. Thanks.
>
>there is something about this in the manual - when looking in the  
>whole brain use cluster or voxel level but only use the corrected p  
>values: either use FWE correction or read the cluster corrected FDR  
>p-value -- using SVC use the voxel level after correction as well ;  
>more on SVC here  
>http://imaging.mrc-cbu.cam.ac.uk/imaging/SmallVolumeCorrection?highlight=(volume)|(correction)|(small)
>
>good luck
>cyril

Thank you for this Cyril. Among the hypotheses, we are proposing that the
drugs act at 4 specific locations, left fusiform, left inferior frontal,
left middle tempporal and left posterior parietal. That's why I was using
"small volume". But I do appreciate the link. I am learning a lot. Thanks. 
- Ananth