[Neuroimaging] 2D NIFTI images in nibabel
pauldmccarthy at gmail.com
Thu Sep 1 12:20:55 EDT 2016
Aah there's my problem - dim0 was set to 2. It's probably `fslroi` that is
the culprit here.
@Robert - I agree with you - this was a 2D slice of a 3D volume, which was
being presented as 2D (due to the dim0 issue). I feel that it should be
presented as 3D volumetric (which it turns out that nibabel correctly does,
if the image header is valid).
Sorry for the noise and thanks,
On 1 September 2016 at 16:56, Christopher J Markiewicz <effigies at bu.edu>
> Can you give us the output of:
> >>> print(nib.load(nifti_file).header._structarr['dim'])
> If the first number is 2, nibabel will (correctly) interpret to mean it
> is to ignore dimensions 3+. If the first number is 3, then there's a bug.
> That said, perhaps it is reasonable that a volumetric image should have
> a minimum of three dimensions?
> On 09/01/2016 05:14 AM, paul mccarthy wrote:
> > Howdy all,
> > Nibabel truncates the third dimension of a NIFTI image with dim3=1.
> > (boo) ws189:nifti2D paulmc$ fslinfo MNI152_T1_2mm_sliceXY.nii.gz
> > data_type INT16
> > dim1 91
> > dim2 109
> > dim3 1
> > dim4 1
> > datatype 4
> > pixdim1 2.000000
> > pixdim2 2.000000
> > pixdim3 2.000000
> > pixdim4 1.000000
> > cal_max 8000.0000
> > cal_min 3000.0000
> > file_type NIFTI-1+
> > (boo) ws189:nifti2D paulmc$ ipython
> > Python 2.7.11 (v2.7.11:6d1b6a68f775, Dec 5 2015, 12:54:16)
> > Type "copyright", "credits" or "license" for more information.
> > IPython 5.1.0 -- An enhanced Interactive Python.
> > ? -> Introduction and overview of IPython's features.
> > %quickref -> Quick reference.
> > help -> Python's own help system.
> > object? -> Details about 'object', use 'object??' for extra details.
> > In : import nibabel as nib
> > In : i = nib.load('MNI152_T1_2mm_sliceXY.nii.gz')
> > In : i.shape
> > Out: (91, 109)
> > In : i.header.get_zooms()
> > Out: (2.0, 2.0)
> > In :
> > Does anybody else think that this is a problem?
> > Note that the dimensions for an image of e.g. size (91, 1, 91) will be
> > preserved.
> > Cheers,
> > Paul
> > _______________________________________________
> > Neuroimaging mailing list
> > Neuroimaging at python.org
> > https://mail.python.org/mailman/listinfo/neuroimaging
> Christopher J Markiewicz
> Ph.D. Candidate, Quantitative Neuroscience Laboratory
> Boston University
> Neuroimaging mailing list
> Neuroimaging at python.org
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Neuroimaging