[Neurodebian-upstream] [Nipy-devel] Fwd: A proposed 64-bit update to the NIfTI format

[Matthew Brett]

Hi,

On Mon, Feb 28, 2011 at 10:52 AM, Valerio Luccio <valerio.luccio at nyu.edu> wrote:
> Yury V. Zaytsev wrote:
>>
>> On Mon, 2011-02-28 at 17:48 +0100, Gael Varoquaux wrote:
>>
>>> Why, oh why, does the neuroimaging community feel a need to invent crappy
>>> format (I know, legacy and good deal of ignorance, just complaining for
>>> the sake of it).
>>
>> Is there a compact explanation somewhere of what this "nifti" is all
>> about and why they can't just use HDF or whatever like everybody else?
>>
> http://nifti.nimh.nih.gov/
>
> First hit in Google ...

Just in general, nifti is the successor to a very old binary file
format for neuroimaging - Analyze.   Analyze (still in use) is a file
with the raw binary image data '<image_name>.img' and a metadata file
'<image_name>.hdr'.  The two most widely used software packages (SPM
and FSL) used slightly different versions of this old format, both
trying to work around the problem that Analyze cannot store a general
relationship between the voxel coordinates and positions in real
space.   Nifti extended analyze by rejigging the 'hdr' to store this
information.  It also specified the standard way of storing the hdr
and img in the same file - a '.nii' file.  There were other changes
but they haven't been widely used yet.  SPM and FSL both use nifti as
standard.   AFNI (another big player) is also standardizing on nifti.

Cheers,

Matthew