Ignacio Vallines wrote:
> Hi Will,
>
> Many thansk for your reply,
>
> My design is actually way too big to use the GUI, so I was using a version of Karl's Batch in
> which I enter the parameters in a piece of code like this (that is why I never saw the option):
>
> SPM.Sess(ii).U(5).P(1).P = abs(RTs1(1:end,2)/1000.);% SPM.Sess(ii).U(5).P(1).i =[1 2];
SPM.Sess(ii).U(5).P(1).h
> = 1;
>
> soooo, you mean just by changing the h field to 2 ???? uhmmm thats is not too complicated...
> nothing like knowing where to look!!!
>
Another idea here is to go thru the GUI for just one subject then save the SPM.mat
Then you can have a look at the data structure that was created and then
edit this appropriately for the other subjects.
> any ideas about intersession averaging???? it another of those things I have bben seraching
> around for a long time... and maybe there is also a simple solution!
>
> many many thanks!!!
>
> Nacho
>
> Institut für experimentelle Psychologie University of Regensburg Universitaetsstrasse 31 93053
> Regensburg, Germany Email: [log in to unmask] Tel. +49 941 943 38
> 49 Fax. +49 941 943 32 33
>
>
>
>>>> Will Penny <[log in to unmask]> 01.10.2004 17:12:48 >>>
>>>>
>
>
> Ignacio Vallines wrote:
>
>
>> Hello,
>>
>> I am a bit confused about what SPM does when you add a non discrete parameter to your events.
>>
>> I have an event related design in which events are entered in the SPM design with a non
>> discrete
>>
> parameter
>
>> value between 0 and 5000 (i.e. reaction times in ms). I have about 40 events per run, so there are
>> virtually no two events with the same parameter (so that no averaging is possible).
>>
>> SPM2 appears to take the parameters to pre-set the height of the HRF, assuming the parameter
>> to behave linearly* and then it just fits the slope to the data* how should I interpret
>> this??? can anyone explain me what it really does???
>>
>> Since I have a short TR of 1.3 secs, my na*ve intention would be to fit one hrf to each event,
>> then plot the beta values for all of the events and then try to find what the optimal fit
>> function to these points is. Does this make sense??? Is there an easy way to do it???
>>
>>
>
> You need to enter the reaction time as a 'parametric modulator'.
>
> After entering the events SPM will prompt you for parametric modulation. Say yes. Then enter the
> reaction time variable (a list of 40 numbers - one for each event).
>
> SPM will prompt for you linear or quadratic modulation - you can go for quadratic here - to test
> for nonlinear modulations of hemodynamic response by reaction time (RT). SPM will then create
> 3 regressor for this event (1) delta functions convolved with an HRF, (2) delta functions whos
> height is multiplied by RT then convolved with an HRF and (3) delta functions whos height is
> multiplied by square of RT the convolved with HRF. You can then use an F-contrast to see if RT
> has an effect on BOLD response (whether linear or otherwise).
>
> Best,
>
> Will.
>
>
>
>> Another problem I am crashing against, is the impossibility of averaging events from different
>>
> sessions
>
>> (data collected on different scan days, for example). Plotting events from different session seems
>> not possible from the GUI, I stepped through the spm_grahp.m code and played with the workspace
>> variables and realized that they where not properly normalized/scaled* is there a trick to
>> improve this???? it seems to be a satandard feature in other packages??? how do people do
>> this in SPM2???
>>
>> I would really appreciate some help on these two issues*
>>
>> Many thanks!
>>
>> Nacho
>>
>>
>>
>> Institut f?r experimentelle Psychologie University of Regensburg Universitaetsstrasse 31 93053
>>
> Regensburg,
>
>> Germany Email: [log in to unmask] Tel. +49 941 943 38 49 Fax. +49 941
>> 943 32 33
>>
>>
>>
>>
>
>
--
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/
|
