Thanks so much!
-Pete
> Hi Peter,
>
>> I've collected resting-state fmri data from a patient group and a
>> control
>> group. After processing the data using the pre-stats tab in FEAT and
>> subsequently running MELODIC, I merged the subjects' melodic outputs
>> together and reran MELODIC. From the resulting output, I've
>> indentified a
>> number of components of interest that I'd like to compare between
>> the two
>> groups.
>
>> My understanding of what I should do is the following:
>>
>> 1) Create a mask representing the combined thresh_zstat maps for
>> both groups.
>
> nothing wrong with this, but you won't need it later
>
>> 2) Concatenate all the thresh_zstat maps of a particular component
>> for all
>> subjects, with all the subjects in one group first, followed by all
>> the
>> subjects in the second.
>>
>
> No, you wouldn't want to run any model based higher-level analysis on
> thresholded lower-level stats images - simply use the associated
> unthresholded maps instead.
>
>
>> 3) Create design.mat and design.con files using the GLM Setup utility,
>> where my design looks like the following:
>> group EV1 EV2
>> groupA groupB Title EV1 EV2
>> s1 1 1 0 | C1 groupA mean 1 0
>> s2 1 1 0 | C2 groupB mean 0 1
>> s3 1 1 0 | C3 A > B 1 -1
>> s4 1 1 0 | C4 B > A -1 1
>> s5 2 0 1 |
>> s6 2 0 1 |
>> s7 2 0 1 |
>> s8 2 0 1 |
>>
>
> That sounds OK
>
>> 4) run a Two-Sample Unpaired T-test using RANDOMISE with the following
>> command:
>>
>> randomise -i <4D_input> -o <output> -d design.mat -t design.con -c
>> 0.99
>>
>
> nope, the argument to the -c (or -C) option is in units of the raw t-
> stats (not z) that is being formed within randomise, i.e. if you want
> a cluster forming threshold of t>2.3 use that number here. The output
> file contains the p-values, i.e. you can threshold that one at 0.99,
> say after having obtained it from randomise.
> hope this helps
> christian
>
>
>
>> Here are my questions:
>> Does it appear I am setting up the design correctly?
>> Is using the option "-c 0.99" the equivalent of looking for
>> significance
>> at p<0.01?
>> Do you have any other suggestions or recommendations for carrying
>> this out?
>>
>> You help is greatly appreciated. Thanks in advance.
>>
>> Cheers,
>>
>> Pete
>>
>> Center for Magnetic Resonance Research
>> University of Minnesota
>
> ____
> Christian F. Beckmann
> University Research Lecturer
> Oxford University Centre for Functional MRI of the Brain (FMRIB)
> John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK.
> [log in to unmask] http://www.fmrib.ox.ac.uk/~beckmann
> tel: +44 1865 222551 fax: +44 1865 222717
>
|