[MINC-users] mincdiff on world values

[Andrew Wood]

Hi all,

I came across some interesting behavior with mincdiff:

$ make_phantom ./phantom.mnc
$ mincmath -add -const 1 ./phantom.mnc ./phantom2.mnc
$ mincdiff phantom.mnc phantom2.mnc
...
>     1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1,
>     1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 ;
67c75,80
<  image-max = 1 ;
---
>  image-max = 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1,
>     1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1,
>     1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
2,
>     2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1,
>     1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1,
>     1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 ;
Binary image comparison:
Images are identical.


mincmath didn't change the image block, but rather modified the header to
change how the voxel values are interpreted, so mincdiff's body comparison
reported equality.

>From a user perspective, it doesn't feel right to see "Images are
identical" for images that a clearly different. I suppose mincdiff is doing
what it advertises, but it might be confusing for some users. I was
wondering if it'd make sense to make mincdiff (optionally) compare *real*
values, since users are typically more interested in those than they are
*voxel* values.

Cheers,
Andrew