Dear John and Luis,
> > Could anyone give me a reference of who came up with the use of a gamma
> > variate function to model the HRF,
>
> Friston (1994) "Analysis of functional MRI time series" HBM: Vol. 1,
> 153-171, implement the gamma curve as an alternative to the Poisson
> (within the convolution framework).
>
> > and the justification?
>
> It looks a little bit like a HRF? -- i don't know of any physical
> justification.
We used a gamma variate because (i) 'It looks a little bit like a
HRF'! and (ii) it represents the next step up from the Poisson model.
The Poisson model has one parameter that effectively couples delay and
dispersion. The gamma model has two parameters that allows delay and
dispaersion to be fitted separately (The Possion is a special case of
the gamma).
With best wishes - Karl
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|