Dear Soren,
You are completely correct that we do not use the most technically
accurate language to refer to these things. That is why the terms
"radiological" and the like always appear in quotes like this in the
FAQ. We use somewhat loose terminology in the hope that it is more
accessible to the less mathematically trained FSL users.
The critical thing is the sign of the determinant of the matrix
that maps voxel coordinates to real-world (or mm) coordinates,
as this determines how an image is treated internally in the code
that reads it from disk. Display is really a completely separate
issue, as you rightly point out.
The big difficulty is that getting the left-right orientation
correct internally in the programs is absolutely critical, and so
people need to check that this is correct in their data. We now
provide labels on the axes in FSLView, and this is really the best
thing to do - just check if these labels are correct or not. The idea
of "radiological storage" is really just a way of stating that the
voxel order stored on the disk is such that the correct matrix to
map these voxel coordinates to real-world coordinates will have
a negative determinant. This kind of technically accurate phrasing
is helpful for some users, but not for many, and is very cumbersome.
I'm very sorry that you have been confused by the term "radiologically
stored" but I had hoped that the FAQ entry entitled
'What exactly is the definition of "radiological convention" or
"neurological convention" used in FSL?'
which included the statement
'Our technical definition of "neurological" convention for image storage
is that the mapping between the voxel-coordinate-system and the
nifti-real-world-coordinate-system (i.e. the sform or qform matrix) has
a positive determinant.'
was sufficiently detailed that it would have made it clearer.
If you have any concrete suggestions on how to improve our FAQ entry
(for the benefit of all users) then please let us know.
All the best,
Mark
On 12 Mar 2007, at 10:20, Soren Christensen wrote:
> Hi,
> I have been fighting with some MINC to nifti converison for the last
> few days and been trying to understand the issues in using
> "neurological" formatted data in FSL.
>
> I noticed that this list often refers to data as stored radiologically
> or neurologically - is this really an appropiate way of putting it?
> As I see it, the pathway from data to screen is this:
> 1) Data is transformed into world (mm) space using specified transform
> 2) The world space is transformed to display using some convention of
> viewing world space be it radiological or neurological. (Whether left
> maps to left or to right on the screen)
>
> As such, radiological or neurological is simply a way world space is
> viewed, in principle, it has nothing to do with the way data is stored
> in the file. If 1) and 2) are mixed together, eg. by a program that
> reads from file directly to the display in a way that depends on the
> file data order then data storage is no longer independent of display.
>
> Is this view correct I have I misunderstood something?
> I hope to not further add to any confusion but the term
> "radiologically stored" confuses me.
>
> Regards
> Soren
|