Hi Sam,
> I've read your 2007 paper and pored through the SPM archives and the
> sense i get is that because my fMRI data was acquired interleaved,
> using it for DCM is suboptimal, even after slice timing correction is
> done as a very first preprocessing step.
> So, is just 'suboptimal' or would using such data make my study total
> crap? I certainly don't want to embark on a study where reviewers
> destroy me based on this technical detail.
I definitely wouldn't throw away the data. It depends on the technical details if and how much suboptimality there is. In general, with a fast overall TR (of the image), slice timing correction of interleaved data will do an appropriate job (i.e. reconstruct the image as if acquired with a non-interleaved sequence), because it only has to interpolate time points not too distant from the actual data points. For example, with a TR of 2 seconds, I personally would rely on the reconstruction. With higher TR, the reconstruction is bound to become less veridical, where I'd be getting slightly worried with TRs > 3 secs, and worried with TRs > 4 secs.
However, it's an empirical and open question whether a suboptimal slice timing correction will make DCM compute results qualitatively different from what you would have got if you had acquired the same brain process with a non-interleaved sequence.
I don't know what reviewers would say. It'd probably be helpful for your cause to identify published DCM papers that are based on interleaved and slice timing corrected data and cite these in support of your analysis technique.
Best wishes, Stefan
P.S.: I hope you don't mind that I copied your email to the SPM list. Your question seemed of general interest to me.
----- Original Message -----
From: "Sam Torrisi" <[log in to unmask]>
To: [log in to unmask]
Sent: Thursday, December 9, 2010 11:43:53 PM
Subject: slice timing correction for dcm
Good day Mr. Kiebel-
I've read your 2007 paper and pored through the SPM archives and the
sense i get is that because my fMRI data was acquired interleaved,
using it for DCM is suboptimal, even after slice timing correction is
done as a very first preprocessing step.
So, is just 'suboptimal' or would using such data make my study total
crap? I certainly don't want to embark on a study where reviewers
destroy me based on this technical detail.
thank you!
-Sam Torrisi
UCLA Mood Disorders Research Program
300 UCLA Medical Plaza, Suite 1544
Los Angeles, CA 90095-6968
|