mriconvert: Surprising behavior when converting motion corrected Siemens Dicom series

[Michael Hanke]

Hi,

I am trying to use mriconvert to convert some Siemens DICOMs into NIfTIs.
The DICOMs
have been motion corrected scanner-side, and each DICOM series has the
reference volume prepended. For an example series MRIConvert reports:

Number of volumes: 451
Number of files: 452

That seems correct. When I iterate over the DICOMs "manually" I find
exactly 452 DICOM files.
The converted NIfTI has 451 -- also correct, but I was curious which volume
was discarded and here it gets a bit strange.

Each DICOM has the motion correction parameters in its ImageComments so I
compared the ones exported by MRIConvert with the ones I extracted
"manually" from each DICOM. For the first image series everything is good:

InstNmbr; "byhand"; MRIConvert
1; Reference volume for motion correction + DiCo Applied;    0.28 0.13
-0.16 0.03 -0.05 0.03
2; Motion: 0.28,0.13,-0.16,0.03,-0.05,0.03 + DiCo Applied;    0.21 0.14
-0.18 0.04 -0.06 0.04
3; Motion: 0.21,0.14,-0.18,0.04,-0.06,0.04 + DiCo Applied;    0.28 0.13
-0.18 0.03 -0.05 0.04
4; Motion: 0.28,0.13,-0.18,0.03,-0.05,0.04 + DiCo Applied;    0.32 0.13
-0.16 0.04 -0.05 0.03
5; Motion: 0.32,0.13,-0.16,0.04,-0.05,0.03 + DiCo Applied;    0.26 0.14
-0.21 0.04 -0.04 0.03

(mriconvert output is truncated and has commas replaced with spaces)

As you can see, MRIConvert discards the reference volume. However for the
second series in the same recording session things are different:

InstNmbr; "byhand"; MRIConvert
1; Reference volume for motion correction. + DiCo Applied;   volume
2; Motion: 0.94,0.11,-1.25,0.16,-0.45,0.24 + DiCo Applied;    0.80 0.12
-1.39 0.14 -0.46 0.25
3; Motion: 0.80,0.12,-1.39,0.14,-0.46,0.25 + DiCo Applied;    1.05 0.11
-1.31 0.14 -0.48 0.26
4; Motion: 1.05,0.11,-1.31,0.14,-0.48,0.26 + DiCo Applied;    0.86 0.13
-1.40 0.17 -0.48 0.24
5; Motion: 0.86,0.13,-1.40,0.17,-0.48,0.24 + DiCo Applied;    1.00 0.11
-1.28 0.16 -0.52 0.27

This time the reference image is left in the series, but the first actual
volume of the recording is
skipped. The total number of volumes in the NIfTI matches the number of
recorded volumes for all series.

Out of 4 datasets that I have, two show this problem in the second
functional run, one in the first, and the last one seems fine. This issue
seems to affect only the start of the timeseries, the rest is in-sync.

I'd be glad if you could shed some light on this issue. Please let me know
if you need more information.

Michael

-- 
Michael Hanke
http://mih.voxindeserto.de
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists-prod.uoregon.edu/pipermail/mriconvert/attachments/20130418/e991973e/attachment-0001.html>