[MINC-users] Display Vi property and float datatype

[Jonathan HARLAP]

Hi Mishkin,

I took a quick look and made a few minor edits, but the section that
seems most needing editing makes some claims that I cannot find basis
for.  Where in the code of Display do you find it dividing by 255 to
do voxel to value conversion?  From what I see, it always (correctly)
uses volume_io's convert_voxel_to_value and convert_value_to_voxel.

Also, you should take into account that not all mincs are native (from
scanner), so your statements about voxel values always being integers
is not strictly correct.  For an easy example, try:

mincmath -double -div <some minc file.mnc> -const 2 test.mnc

You'll then see voxel values that step by 0.5 rather than 1.
Obviously this is a very artificial example, but probability maps
stored as MINCs will likely exhibit this behaviour.

Out of curiousity, which Alex do you refer to in your page?

Cheers,
J

On Tue, Feb 20, 2007 at 02:05:09PM -0500, Mishkin DERAKHSHAN wrote:
> I've been trying to wrap my head around the Display Vl problem for quite
> some time. I wrote my explanation of how Vx and Vl are calculated in
> Display versus how they are calculated in volume_io, but it would be great
> if someone who truly understood what was going on could edit the page.
> 
> http://wiki.bic.mni.mcgill.ca/index.php/RealVsVoxelValue
> 
> mishkin
> 
> On Sat, 27 Jan 2007, Jason Lerch wrote:
> 
> > Display (and register) load files as byte files. So the only way to
> > work around this problem is to convert your volume to byte first -
> > then the values will be consistent across Display and whatever other
> > program you are using.
> >
> > Jason
> >
> > On 27-Jan-07, at 9:11 PM, Soren Christensen wrote:
> >
> > > Hi,
> > >  I have a MINC file of floats with discrete values 0-2-4-6 etc.
> > > When viewed in Display the values appear as 0-1.904 - 4.024 - 5.92
> > > etc.
> > > The actual values in the file appear to be ok. as they are 2-4-6
> > > when loaded
> > > into Matlab via EMMA, and also mincmath can identify the values of
> > > 2 (and
> > > not 1.904) - so it seems to be related to Display and not the
> > > actual file
> > > content.
> > > Is there a way to change how Display calculates the Vi values?
> > > It is a problem since I'd like to set a threshold for ROI
> > > segmentation in
> > > Display.
> > >
> > > Thanks,
> > > Soren
> > > _______________________________________________
> > > MINC-users at bic.mni.mcgill.ca
> > > http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users
> >
> > _______________________________________________
> > MINC-users at bic.mni.mcgill.ca
> > http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users
> >
> _______________________________________________
> MINC-users at bic.mni.mcgill.ca
> http://www.bic.mni.mcgill.ca/mailman/listinfo/minc-users