Hi,
A general comment think carefully about the timing: depending on your TR and when the slices of interest are acquired relative to the onset of the stimulus and start of the scan volume the temporal derivative will be important in for accounting for timing differences. Another is put all your knowledge about your stimulus in the same model (different regressor for short and long). You could consider trying slice timing correction and a fixed canonical HRF (no derivatives) in this instance, because timing is important and your stimulus might not allow for you to account for this in the recommended way (with derivatives).
You probably really need to think about the study design and the timing of the stimulus relative to slice acquisition (i.e. vary it). A task where some sustained sounds did not have a word would mean you could do a subtraction of sound with/without the short word stimulus imbedded and get at your effect?
David Carmichael
Department of Clinical and Experimental Epilepsy
UCL Institute of Neurology
Correspondance to
AMRIG
12 Queen Square
London, UK
WC1N 3AR
Tel 02076762006
Fax 02076762005
-----Original Message-----
From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On Behalf Of Michel Hoen
Sent: 08 September 2010 16:26
To: [log in to unmask]
Subject: [SPM] Bold response dynamics and time-serie modelling
Dear SPM experts,
I'm analysing data from a study where subjects have to listen to one word presented in background noise.
I was trying to model two types of events, the raw presentation of the noise (one "sustained" event lasting 3.5 s) and the presentation of the word embedded inside that noise (one "short" event lasting 1s).
Yes I am looking for this needle in that haystack.
If I model the short event only, will the correct form of the BOLD response be the HRF, and anyway what will it reflect lots of hay and a bit of needle ?
In other words, if you first stimulate sensory areas with a sustained stimulation and add on top of that a transient event, will the BOLD response to that transient event still be the canonical HRF ??
I'm asking that because actually we have very nice responses on the time-derivative but not on the canonical HRF and I was wondering if eventually the special combination of stimuli: 1 sustained + 1 transient on top (or deeply hidden in it actually ....) would modify the form of the bold response to the transient stimulation to something different than can_HRF.
Any remarks, thoughts, references, I guess this should be the same for any type of paradigms using 1 sustained + 1 transient stimulation in the same modality.
Thanks very much in advance,
Michel Hoen
|