[Insight-developers] Bug changing volume metadata by looking at it

[Bradley Lowekamp]

Hello,

We are experiencing a rather critical bug, that is changing our Origin, when viewing the "Volume Information" in the Volumes module. This is on yesterdays nightly. I don't know when it started to occur, as we are starting to look at some older dataset which were previously segmented.

This sounded similar to some issues mentioned during the TCON related to the double spin boxes or precision preferences.

Initially:

>>> lm1 = sitkUtils.PullFromSlicer("A Close")
>>> print lm1

Image (0x13b2ca030)
  RTTI typeinfo:   itk::Image<unsigned short, 3u>
  Reference Count: 1
  Modified Time: 3771
  Debug: Off
  Observers: 
    none
  Source: (none)
  Source output name: (none)
  Release Data: Off
  Data Released: False
  Global Release Data: Off
  PipelineMTime: 3597
  UpdateMTime: 3770
  RealTimeStamp: 0 seconds 
  LargestPossibleRegion: 
    Dimension: 3
    Index: [0, 0, 0]
    Size: [2618, 1368, 325]
  BufferedRegion: 
    Dimension: 3
    Index: [0, 0, 0]
    Size: [2618, 1368, 325]
  RequestedRegion: 
    Dimension: 3
    Index: [0, 0, 0]
    Size: [2618, 1368, 325]
  Spacing: [160, 160, 480]
  Origin: [547339, 218860, 20904.4]
  Direction: 
-1 0 0
0 -1 0
0 0 1

After viewing in Volumes module:
>>> lm1 = sitkUtils.PullFromSlicer("A Close")
>>> print lm1
Image (0x13b222ad0)
  RTTI typeinfo:   itk::Image<unsigned short, 3u>
  Reference Count: 1
  Modified Time: 4481
  Debug: Off
  Observers: 
    none
  Source: (none)
  Source output name: (none)
  Release Data: Off
  Data Released: False
  Global Release Data: Off
  PipelineMTime: 4307
  UpdateMTime: 4480
  RealTimeStamp: 0 seconds 
  LargestPossibleRegion: 
    Dimension: 3
    Index: [0, 0, 0]
    Size: [2618, 1368, 325]
  BufferedRegion: 
    Dimension: 3
    Index: [0, 0, 0]
    Size: [2618, 1368, 325]
  RequestedRegion: 
    Dimension: 3
    Index: [0, 0, 0]
    Size: [2618, 1368, 325]
  Spacing: [160, 160, 480]
  Origin: [10000, 10000, 10000]
  Direction: 
-1 0 0
0 -1 0
0 0 1

Note the change in the origin! We have also observed other odd behaviors such as ROI changing, location, but it's hard for me to figure how what has change with respect to where in those cases.

This data set is a little odd of us because the user specified the units in angstroms and not nano-meters, so everything is 10X bigger than normal.

I will continue to investigate and will write a test which uses synthetic data to reproduce the above, hopefully without needed so many pixels.

Thanks,
Brad