[FieldTrip] Segmentation Memory Issue

Jesse Tinker jessetinker at gmail.com
Fri May 17 21:36:12 CEST 2013


Hello Fieldtrippers!

This is my first post ever and I just want to say thank you to everyone for
promoting such a vibrant and supportive intellectual community.

I have an issue which seems so simple but I am somewhat stymied by: using
fieldtrip to accomplish the coregistration. The main points of my code are:

mri = ft_read_mri( );

mri = ft_volumerealign(cfg, mri);           % Mark fiducials

mrireslice = ft_volumereslice(cfg,mri);  %To apply the resultant
transformation matrix

segmentedmri  = ft_volumesegment(cfg, mrireslice);



First, if I skip the reslice step, everything works fine. However, if I do
the reslice and then try to segment, spm_preproc tries to create a 1.8 Tb
array. Although it bravely tries, my 24 Gb of RAM inevitably fails. The
analogous array for the unsliced mri is 2Mb.

I found this confusing as the quantity of information in the resliced image
is the exact same. I traced through all the coding and the computation that
fails is based on the transformation matrix of the MRI which, of course,
changes with the reslicing. I am happy to provide details of that but won't
at the moment.

My question, has anyone else encountered this or developed a workaround?

Very much appreciated,

Jesse
British Columbia Mental Health and Research Institute
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20130517/65a54bb0/attachment.html>


More information about the fieldtrip mailing list