Hi,
There is no software reason that you cannot use both.
If you are concerned that slice timing correction degrades the data quality, because of the
interpolation effects, then there is some reason to avoid it. The big difference between the
two is that the temporal derivatives have the capacity to shift the timing differently to the
exact slice timing specifications, which allows it to soak up differences due to HRF timing
differences (from the canonical HRF).
There is no clear opinion on which method or combination of methods is "best" - but
certainly the combination is a reasonable option. The result will depend a lot on the
sensitivity of the experiment to exact timing, and hence is more of an issue with longer
TRs and event-related designs. There is also an issue of correlation between temporal
derivatives in closely packed event-related designs, and this complicates things a little more.
So, the summary is that there is no simple summary, but it is fine to consider using both
slice timing and temporal derivatives as a valid processing option.
All the best,
Mark
On 2 Feb 2012, at 19:53, Cinly Ooi wrote:
> Dear Everyone,
>
> In 1st level FEAT, I normally choose the Slice Timing Correction _and_ EV Full Model Setup's "Add Temporal Derivative".
>
> However, when discussing it with another researcher, he is of the opinion that we should use either one and not both. In fact, he says it is better to use "Add Temporal Derivative". After the discussion, I went back to check FSL webpages and still cannot see why we cannot use both. I am interested to find out what do you think.
>
> From my point of view, both aims to fit the data to the waveform better in its own way. Slice Timing by 'repositioning' the slice (using Hanning-windowed interpolation) and Temporal Derivative works by 'softening' the harsh waveform so that it stand a better change of a better waveform in the presence of timing error and other noise. Both are complementary. If I use both, at worst what I get is one of the step is redundant. However, I acknowledge that I am simply looking at this from 10,000 ft and the real life nitty-gritty of statistics and computing will no doubt complicate the matter.
>
> Thanking you all in advance and Best Regards,
> Cinly
>
|