Dear Jeng-Ren
> We are now trying the fMRI analysis with event-related module. There are many problems occurred during the process. We have designed a paradigm with single trial (one trial type only) according to the design in the paper of Josephs (1997). The interscan interval is 2 secs, there are 15 scans during one trial, and the trial is repeated for 8 times.
The first thing to note in the efMRI module in SPM is that event times have to be entered in TRs or time between scans. E.g. if you have a TR of 4 seconds and your first event happened at 10 seconds, you would specify 2.5 in SPM.
> We tried to analyze the data with event-related module of Fourier sequences harmonics, but we did not know how to set the parameter for every field appropriately.
In general SPM allows the responses to be modelled by a variety of basis functions. If the time between each event is always the same you can use a Fourier set to analyse your data, because the sines and cosines are ideal to model the periodic structure. If the events are non-periodic (inter-trial intervals vary) you can still use a short, event-locked Fourier set. More parsimonious models are the canonical HRF with its time and dispersion derivative.
Which model for what ?
If you are interested in simple responses to brief stimuli, HRF (+ temporal derivative) might be enough
If you expect differences in the shape of the HRF, e.g. working memory with varying load and time to keep things in working memory, you want to use a more flexible model, ie. 3 gamma functions + derivatives or the Fourier set.
> Can you give us an example for the parameter setting when we use the Fourier sequences harmonics? In addition, what is the physical meanings of the basis function of Fourier sequences harmonic analogized to the brain system?
The basis function of a Fourier series are just sines and cosines.
-Christian
--
Dr Christian Büchel
Wellcome Department of Cognitive Neurology
Functional Imaging Laboratory
12 Queen Square, London WC1N 3BG, UK
Tel.: +44-171-833-7483
Fax.: +44-171-813-1420
email: [log in to unmask]
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|