Tom,
According to the example that you've gave me my design matrix should look
like that:
Group EV1 EV2 ..........EV43
> 1 1 0 0 1 0 0 0 0 0
> 1 -1 0 0 1 0 0 0 0 0
> 2 1 0 0 0 1 0 0 0 0
> 2 -1 0 0 0 1 0 0 0 0
> ...
> 40 0 1 0 0 0 0 0 1 0
> 40 0 -1 0 0 0 0 0 1 0
> 41 0 1 0 0 0 0 0 0 1
> 41 0 -1 0 0 0 0 0 0 1
Now my groups are not only 2 but 41coupled (since my data set consists of 41
pairs of patients devided in 23 Siemens scanned pairs and 18 GE scanned
pairs.the same holds for the controls).The way that i interpreted the
previous design matrix is the the following:
Number of groups :82 (41 couples)
Number of EV's: 43
The first two EV's express the difference between The Siemens Group
(controls-patients) and the GE group (controls-patients)
The other 41 Ev's are one for each subject pair (Siemens control-Siemens
patient and GE control-GE patient)
Group EV1 EV2 ... EV42 EV43
1 1 0 (Siemens Control) 0 0
1 -1 0 (Siemens Patient) 0 0
2 1 0 (Siemens Control) 0 0
2 -1 0 (Siemens Patient) 0 0
...
23 1 0 (Siemens Control) . 0 0
23 -1 0 (Siemens Patient) . 0 0
24 0 1 (GE Control) . 1 0
24 0 -1 (GE Patient) 1 0
...
41 0 1 (GE Control) 0 1
41 0 -1 (GE Patient) 0 1
Is this interpretation correct?or did i misunderstand something?
If this interpretation holds then i'm quite confused on how it will be
possible to extract information concerning the two groups Patients and
Controls?Cause instead of having 2 groups now we have 41 and the grouping is
among Siemens Scans and GE scans instead of Controls and Patients.Besides
this is what EV1 and EV2 express(difference between Siemens group and GE
group).Right?
In the case of 2 groups (patients and controls), you just need 4 contrasts
(A-B,B-A,A,B) to extract this info but how about my case?
I would really appreciate your help once more cause i'm stuck here.
Thank you again
Antonios-Constantine Thanellas
On Fri, 11 Apr 2008 11:55:53 +0100, Thomas Nichols <[log in to unmask]>
wrote:
>Antonios,
>
>This design would be a 'two group, paired difference'. Your design matrix
>should look like the Feat " (Two-Sample Paired T-Test)" design *but* with
>two columns for the paired differences.
>
>Using a slightly different order than the Feat web page (i.e. grouping
>subjects together) the design matrix will look like
> 1 0 0 1 0 0 0 0 0
>-1 0 0 1 0 0 0 0 0
> 1 0 0 0 1 0 0 0 0
>-1 0 0 0 1 0 0 0 0
>...
> 0 1 0 0 0 0 0 1 0
> 0 -1 0 0 0 0 0 1 0
> 0 1 0 0 0 0 0 0 1
> 0 -1 0 0 0 0 0 0 1
>but then you *also* need to define a group file which tells randomise about
>the pairing
>1
>1
>2
>2
>3
>3
>...
>That should do it.
>
>-Tom
>
>
>On Thu, Apr 10, 2008 at 3:43 PM, Antonios - Constantine Thanellas <
>[log in to unmask]> wrote:
>
>> Dear fsl users,
>>
>> I have 30 T1 pairs (baseline and follow up) of controls scanned with
>> Siemens
>> scanner and 21 T1 pairs scanned with GE scanner
>>
>> My patients consist of 23 pairs of Siemens scanner and 18 pairs of GE
>> scanner
>>
>> After longitudinal analysis i want to continue with the use of randomise
>> and
>> localize differences between controls and patients.
>>
>> I supose that the first step is to use grouped matched controls and
>> patients
>> (23 Siemens and 18 GE controls as one group and the second group 23
>> Siemens
>> and 18 GE patients). Is my assumption correct?
>>
>> I'm quite confused on how am i gonna set up the contrasts and the design
>> matrix. Should i consider my case as an Unpaired Two-Group Difference
>> (Two-Sample Unpaired T-Test) and set up my contrast files as it is
>> mentioned
>> in the corresponding part of the help page of FEAT or my case falls into
>> the
>> case of Paired Two-Group Difference (Two-Sample Paired T-Test)?
>>
>> Thanks in advance
>> Antonios-Constantine Thanellas
>>
>>
>
>
>--
>____________________________________________
>Thomas Nichols, PhD
>Director, Modelling & Genetics
>GlaxoSmithKline Clinical Imaging Centre
>
>Senior Research Fellow
>Oxford University FMRIB Centre
>
|