Dear Karl,
that's a challenging design ...
You do not really have any continuity in your time series, which would allow a sufficient modelling with an HRF function. Using a FIR model instead might work. Handling it as XX independent sessions, as Simone suggested, might work as well, but will give you a very low sensitivity.
Another problem that I see is that you are not having a good baseline. The only 'true' baseline, the fixation, have you to throw away, since it contains the T1 effect. The 'Instruction' could perhaps work as baseline, but will already create some activation. Does this task need this 7 scans or are the subject finished with it earlier? Than you try to model it as a short, single event with a decay period of the BOLD signal.
When the task takes almost all of these 7 scans, you are missing this additional information from the decay of the BOLD signal, which could last up to 20 seconds after the neuronal activation has stopped. This could give you valuable information for modeling your signal. On the other hand, what you may pic up with your design is the decay of the BOLD signal from the oral response, since you are not waiting until this effect is gone.
So, apart from revising the design, a model with XX sessions and a HRF model might work, when the actual performance of the task does not last the full 7 scans, or a FIR model with concatenated scans, but both ways are not very optimal.
Good luck,
Karsten
> Dear experts,
>
> We wonder whether you can give us a clue on how to handle the analysis of the following fMRI design:
>
>
> 1 block consists of (1 scan = TR = 2sec)
>
> - 2 scans: fixation
>
> - 2 scans: trial instruction
>
> - 7 scans: working on the task
>
> - Then we stop the scanner for 10 seconds to record the oral responses
>
> After each block, the next block starts with the fixation.
>
>
>
>
> There are 4 conditions and we present every condition 8 times.
>
>
>
> Because of different steady state effects we throw away the first two scans (fixation). What we are interested in is the activation related to working on the task. We think the best way is to give SPM the onset of this working task and use an event-related design.
>
> The problem is we do not know how SPM handles such scanner breaks.
>
>
>
> Thanks a lot in advance for your help.
>
>
>
> Best regards,
>
> Roland & Karl
>
>
>
> ETH Zurich
>
> Dr. Roland H. Grabner
>
> Research on Learning and Instruction
>
> Institute for Behavioral Sciences
>
> Universitaetsstrasse 6, CAB G 83.2
>
> CH-8092 Zurich, Switzerland
>
>
>
> [log in to unmask]
>
> http://www.ifvll.ethz.ch
>
>
>
> +41 44 632 50 53 phone
>
> +41 44 632 12 19 fax
>
>
--
-------------------------------------------------------------
Karsten Specht, PhD
Department of Biological and Medical Psychology
& National Competence Centre for functional MRI
University of Bergen
Jonas Lies vei 91
5009 Bergen
Norway
Tel.: +47-555-86279
Fax: +47-555-89872
[log in to unmask]
http://fmri.uib.no/
|
