Dear Ping,
I'm not convinced this is a good experimental design.
A better experiment would be something more specific eg.
the effect of drug on a particular task. Say task T.
This would excite whatever brain systems/processes you are studying ..
Then you could have a boxcar regressor which is `on'
when the animal is performing the task and `off' otherwise
with a period of, say, 30s.
You'd have one boxcar regressor for each session and
then look at a [1 0 -1 0] F-contrast (spanning the
regressors T in session 1, mean in session 1,
T in session 2, mean in session 2) to look for drug-related
changes.
Best,
Will.
Wang Ping wrote:
> Dear William,
>
> I just have another problem:
>
> In following experiment, how to design the matrix?
> Suppose in the fMRI experiment, the subject is animal, we study the
> drug's effect in the animal brain, one subject, one session.
> First we scan the anmial for several minutes as the baseline, then begin
> to inject the drug to the animal (the scanning is going on), after
> injection, the scan will keep some time. I don't know for this case,
> how to perfrom the design matrix?
>
> Sincerely yours,
> Ping
>
>
>
>> From: Will Penny <[log in to unmask]>
>> Reply-To: Will Penny <[log in to unmask]>
>> To: [log in to unmask]
>> Subject: Re: [SPM] what's the meaning in the desian matrix based on SPM2?
>> Date: Thu, 15 Sep 2005 12:12:34 +0100
>>
>> Wang Ping wrote:
>>
>> > Dear spmer,
>> >
>> > I just try to use spm2 to design matrix, and finally in the figure
>> > window get the design matrix. I want to know what's the meaning of
>> > something like "sn(1) trial 1*bf(1)"? I don't know the meanings of
>> > sn(1) and bf(1),
>>
>>
>>
>> You can put data from multiple sessions into the same design
>> matrix. sn(1) means session number 1.
>>
>> trial 1 is the first trial in that session. If you've specified
>> 4 different conditions (trial types) this index will run from
>> 1 to 4.
>>
>> bf(1) is the 1st basis function. Each trial can be modelled with
>> a number of basis functions eg (1) canonical HRF, (2) temporal derivative
>> (3) dispersion derivative.
>>
>>
>> I also don't know the design orthogonality.
>>
>> This tells you how much correlation there is between regressors.
>>
>> If you have two vectors, eg columns in your design matrix, say
>> x1 and x2 then the inner product between them
>>
>> x1^T x2 = |x1| |x2| cos theta
>>
>> where |x1| and |x2| are the lengths of x1 and x2 and theta
>> is the angle between the vectors.
>>
>> See section 2.5 of
>>
>> http://www.fil.ion.ucl.ac.uk/~wpenny/course/matrices.ps
>>
>> for derivation and discussion.
>>
>> Two vectors are orthogonal if the inner product is 0.
>> This will be the case if cos theta is 0.
>>
>> In SPM, if you go to Review->Design->Design matrix SPM
>> will plot a matrix of cos theta's where the i,j th
>> entry corresponds to the value for columns i and j.
>>
>> Generally, experimental designs are better if your
>> columns are more orthogonal.
>>
>> Best,
>>
>> Will.
>>
>>
>> Could
>>
>> > anyone tell me?
>> >
>> > Thanks so much!
>> > Ping Wang
>> >
>> >
>>
>> --
>> William D. Penny
>> Wellcome Department of Imaging Neuroscience
>> University College London
>> 12 Queen Square
>> London WC1N 3BG
>>
>> Tel: 020 7833 7475
>> FAX: 020 7813 1420
>> Email: [log in to unmask]
>> URL: http://www.fil.ion.ucl.ac.uk/~wpenny/
>
>
>
>
>
--
William D. Penny
Wellcome Department of Imaging Neuroscience
University College London
12 Queen Square
London WC1N 3BG
Tel: 020 7833 7475
FAX: 020 7813 1420
Email: [log in to unmask]
URL: http://www.fil.ion.ucl.ac.uk/~wpenny/
|