Dear Leonhard,
On Mon, 26 Mar 2007, Leonhard Schilbach wrote:
> we are interested in using an FIR approach for the re-analysis of a
> previously acquired event-related design. As I am unfamiliar with this
> method, I would be very grateful if someone could point out relevant
> literature pertaining to the underlying ideas and assumptions.
The finite impulse response (FIR) model is widely used in signal
processing and most books in that field would contain information about
the model. It is linear (both wrt. input and parameters) and
time-invariant. In some of the cases I have seen it tended to overfit,
making an "un-physiological" shape for the hemodynamic response. Whether
this is important for testing with the general linear model I do not know.
Apart from the references mentioned (e.g., Ollinger) we have a paper that
comments on and describes the FIR model and suggests a somewhat elaborate
estimation of the parameters in the model: "Modeling the hemodynamic
response in fMRI using smooth FIR filters"
http://dx.doi.org/10.1109/42.897811
The FIR model is also described in "Introduction to Statistical Parametric
Mapping" http://www.fil.ion.ucl.ac.uk/spm/doc/intro/
And Henson compared the the canonical hemodynamic response with the FIR
filter: http://www.fil.ion.ucl.ac.uk/spm/doc/papers/rnah_choice.pdf
sincerely
Finn Årup Nielsen
____________________________________________________________________
Finn Aarup Nielsen, IMM DTU, Denmark
Lundbeck Foundation Center for Intergrated Molecular Brain Imaging
http://www.imm.dtu.dk/~fn/ http://nru.dk/staff/fnielsen/
____________________________________________________________________
|