[MINC-users] mnc2nii conversion problem

[Andrew Janke]

> > --> the result: before conversion values (before -> after) are 654 -> 672 in one point. If i take a different points its for example 232 -> 116 and so on. There is no real pattern but if one looks at the images with register both seems to be "quite" similar.
> >
> > Any suggestions what i did wrong?
>
> This looks like the way minc stores integer data with ranges. Try -float instead
> of -short. Is this better?

Claude is correct, there is always going to be rescaling going on with
short/byte/int MINC images unless you pull a bunch of tricks with the
nii2mnc and mnc2nii conversion. Given that you seem to want to
preserve the real values in your data and not the voxel values what
Claude is suggesting (using float) is the best alternative.

a simple:

   mincreshape -float short.mnc out.mnc

before the conversion to nii should solve the problem.

> > P.S.: I tried to install minc 2.0.13 and 2.0.14 with the following error when i type "make" after ./configure
>
> You have to tell configure that you are building minc2. Try the following:
>
> ./configure --enable-minc2 --with-minc2 <other options here>
>
> I think one of the two minc2 flags is obsolete, but it does work with both. You
> also need --with-build-path=... for the path of hdf5 and netcdf.

The correct configure flag for minc2 itself is --enable-minc2 and for
the other programs when linked against minc2 is --with-minc2.  I dont
agree with the ambiguity but I didnt write that bit! (this is my
excuse)  I subsequently added --with-minc2 to the minc library build
to save on some confusion and left the old --enable-minc2 flag in
there for backwards compatibility.



a