[MINC-users] crash related to a long history

[Pierre Bellec]

Dear minc-users,

I just ran into the following bug when trying to run the Feb-14-2008
quarantine of CIVET on a T1 minc volume. NU_CORRECT crashed with the
following error :
spawn: exec of minc_modify_header failed: Liste d'arguments trop longue

I checked the history of the file, and found that there was a single
dcm2minc command applied to ... over 3000 files ! Basically the history
looked like :
--- History of subject_mri.mnc ---
[01] Fri Apr 16 17:00:55 2010>>> /usr/local/bic/bin/dcm2mnc \
     /data/dicom/dicom-XXXX-subject_20091004.151045_15_1.dcm \
     /data/dicom/new \
     /data/output_men/
with XXXX going from 0000 to 1993, with every number up to 1220 or so
associated to two files (that rule stopped at something like 1222 which by
itself is a bit puzzling). Note that I modified the path and file name to
remove the name of the study/subject.

I cleared the history by doing :
$ minc_modify_header -sinsert :history='' subject_20091005_151045_16_mri.mnc
and that solved the problem. Voila.

Now I don't know if the "bug" can be attributed to the way dcm2mnc was
invoked (I didn't make the conversion myself so I can't comment much on
that), a limitation of NU_CORRECT or a limitation of the way CIVET modifies
the history. In any case I thought this issue might be of interest for the
MINC users community.

Best regards,

Pierre Bellec, PhD
Chercheur post-doctorant
Centre de recherche de l'institut de Gériatrie de Montréal
4565, Chemin Queen-Mary
Montréal (Québec)
H3W 1W5
http://wiki.bic.mni.mcgill.ca/index.php/PierreBellec
tel:       (001)(514) 340 3540 #3367
fax:      (001)(514) 340 3530