Hi,
There's no simple answer to this question.
The approximation gets progressively worse as the shift increases, and it depends on the
TR as to how quickly it gets "bad". It will also depend on what your design is like as to how
important the approximation errors are.
It is normally quite reasonable to expect it to compensate for the majority of the differences
in timing (due to slice timings and variations in the HRF) with a TR of 2-3 seconds.
If you really want to quantify the accuracy more, then you can simulate some delays and
run FEAT with a temporal derivative to see how well it compensates for these with your
specific TR and design.
All the best,
Mark
On 19 Feb 2012, at 18:44, Gabor Perlaki wrote:
> Hi all,
>
> I've read in several articles that temporal derivative can compensate for a "small" temporal shift between the simulated and the actual Bold-responses. But, how much is the maximum temporal shift which can be compensated by this way? Could anybody help me in this issue?
>
> Thanks, Gabor
>
|