Dear SPM masters, slice timers, Darren (& Michael: where has he been?),
in particular:
I am still troubled by the slice timing. Here is an example:
A time series of volumes - each volume of 34 slices is acquired
interleaved in 3.494 s, the next measurement is taken after a delay of
0.256 s. The delay shall be necessary for external triggering in order
to prevent the trigger to hit some sort of "partially refractory" period
of accepting external signal inputs. Thus, the interscan intervall ("TR"
/ "RT") amounts to 3.750 s, and I specify this in the slice timing.
Right?
However, the acquisition time ("TA") is such situation is definitely not
3.6397 s as suggested by SPM99. Obviously, SPMīs routine thinks of its
"timing(2) as the time between last slices and next" in terms of the
beginning of the final slice of a volume and the beginning of the first
slice of the next volume in the time series. In my example, I should
therefore specify TA with 3.391 s = 3.494ms * (1 - 1/nslices). Is that
right?
Interleaved at our scanner means 1 3 5...33 2 4 6...34. The slices are
written to a mosaic in that very order, i.e. not anatomically but
chronologically correct. However, in Siemenīs convention 1 is top. After
converting my slices to analyze volumes, SPM displays the slice with a
z-voxel value of 34 at top and 1 at the bottom. To specify slice order,
I therefore decided to take: 34 32 30...2 33 31 29...1 Right? Finally, I
decided to use the first slice aquired (-> 34) as the reference. I am
not sure if my proceeding is ok - however, the slice timing would seem
to run ok. But where can I check if it indeed did introduce the proper
corrections and how do I do that?
Darren has written a modification to include chronological order to the
process of slice timing correction (spm_slice_timing, Date: Mon, 26 Jun
2000 23:50:35-0500 in the archive). I have not yet used it and wonder if
it would be nice in such an example. I guess my question is: What will I
get from it in my example, Darren?
Finally- when I change the defaults in spm_defaults.m to fMRI_T=34 and
fMRI-T=34, SPM will generate design matrices with time bins of TR / 34.
In my example, it will assume time bins of 110 ms (3.75 s / 34).
However, how does that go with the extra delay of 0.256 s in my example?
Any help would be greatly appreciated. I think many would welcome some
clarification, and I did not mean to bore people with my example but I
felt it includes the points where things may go wrong. TIA- andreas
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|