> Yes 84 million voxels, I don't know why massive this image, but we have
> got a raw DICOM data from 3T Philips MRI scanner for 3D volumetry
> without gap.
Well, in that case you do not want to realign: realign is for time
series where image 2:n are aligned to image 1 (or a mean image). I
figure what you want is to normalize the images so you can compare them.
> So I try to change small dimension size but I cannot do that. I gain a
> confident belief this dimension, probably...so it happen to error, Do
> you know way that reduced dimension size ?
You can try a script called resize_img.m as available from
(the list archives are quite helpful if you are looking for answers) but
if you try to normalize directly, things may work. If not (as seemed to
be the case from your error message) you could try to convert the images
to another format (but loose information), use a less extensive
normalization scheme (it worked until iteration 16) or buy more RAM.
This may actually be the most worthy investment in the long run.
Marko Wilke (Dr.med./M.D.)
[log in to unmask]
Universita"ts-Kinderklinik University Children's Hospital
Abt. III (Neuropa"diatrie) Dept. III (Pediatric neurology)
Hoppe-Seyler-Str. 1, D - 72076 Tu"bingen
Tel.: (+49) 07071 29-83416 Fax: (+49) 07071 29-5473