Hi!
I am analyzing datasets which contain 580 timepoints (TR 3.0 s). There are
four events at volumes 20, 140, 300, and 440. If I have a relatively long
period of time that I want to model with FIR set (140 timepoints after each
event with TR 3.0), should I reduce the number of basis functions from 140.
This high number seems to result in a infinetely long generation time for
the design matrix. I wonder what might be a more suitable number?
Further, I would like ask some advice about using the FIR modelling. Should
the basic waveform in my case or in general be a squarewave with on
duration of the length of TR for example and with off duration same as event
interval?
In the FIR basis functions the window parameter seems to determine the time
interval which is modelled with FIR set and Number parameter then seems to
determine the number of impulse responses used to model this interval. Have
I understood this correctly?
--
Antti Korvenoja
Biomedicum Helsinki, Radiology
P.O. Box 700, 00029 HUS
Finland
tel: +358-9-47171786 : mobile: +358-40-7498469
|