Discrepancy in Hounsfield Units displayed between v2.1.0 and v3.2.0 in ITK-Snap

[Sarah Schmiedel]

When I open the same MHA file in v2.1 and v3.0 of ITK-Snap and navigate to the same (x,y,z) coordinate, the hounsfield units or intensity field is different. Here are some steps to reproduce, thank you for your help! 

1. Load dataset into ITK-Snap v2.1.0.
   
2. Select Tools, Image Info... 
   
3. Enter the coordinate values in the x, y, and z fields
   
4. Record the HU value within the Intensity field under "Voxel Under the Cursor".
5. Load the same dataset from step 1 into ITK-Snap v3.2.0
6. Select the Cursor Inspector icon in the toolbar UI and enter the same coordinate values into the x, y, and z fields as step 3. 
   
7. Observe that the value displayed in the intensity field under "Intensity under cursor" in the toolbar UI is a different Hounsfield Unit value than recorded in step 4.
   

Thanks again for your help! 

[Paul Yushkevich]

Hi Sarah

Would you mind sending a screenshot to help see the difference?

Thanks!
Paul

--
You received this message because you are subscribed to the Google Groups "itksnap-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to itksnap-user...@googlegroups.com.
To post to this group, send email to itksna...@googlegroups.com.
Visit this group at http://groups.google.com/group/itksnap-users.
For more options, visit https://groups.google.com/d/optout.

--

Paul A. Yushkevich, Ph.D.
Associate Professor
Penn Image Computing and Science Laboratory
Department of Radiology
University of Pennsylvania

[Sarah Schmiedel]

Hi Paul, 

On the left, ITK-snap v3.2, the intensity is labeled as 143 at the coordinates (200, 200, 200)

On the right, ITK-snap v2.1, the intensity is labeled as 211 at the same coordinates. 

Both ITK-snap versions have the same data loaded in the same format (medical image, .mha format). 

Thank you for your help! 

Sarah 

[Paul Yushkevich]

Hi Sarah,

This is very odd. You said you are using version 2.1 of ITK-SNAP, and the user interface looks like 2.x, but the window title says 3.0.0-dev. 

Do you know which is the correct value for this voxel?

If you could send the MHA file, that would be helpful too.

Thanks!
Paul

[Sarah Schmiedel]

Hi Paul, 

It is very strange, but I was also able to reproduce using v3.2 compared to v2.1. See the screenshots attached. 

I do not know the correct value for this voxel. 

Any medical image data will work and show this discrepancy. For the screenshots attached, I used the DICOM sample downloaded here: http://www.osirix-viewer.com/datasets/ 

The package with Alias name: Ferovix

Thank you!

Sarah

[Paul Yushkevich]

Hi Sarah,

The difference between the versions is the indexing. The new version uses voxel indices between 1 and K, where K is the number of slices, so between (1,1,1) and (308,512,308) for the image in question. This is more familiar to users of MATLAB, etc.

The old version is more inconsistent. It uses zero-based indices like in C in the layer inspector dialog, but in the slice views itself, it reports one-based indices. So in your screenshot, the dialog has 200/200/200 but the slice views say "Slice 201 of 512", etc. 

If you set the index in the old ITK-SNAP to 199/199/199 you will see 336, same as in the new version. You can also see that the world coordinates (NIFTI) on the layer inspector agree between the two versions then.

I realize this can cause confusion, but I think this was a problem in the old version that the new version fixes - so it made sense.

Thanks!
Paul

[Sarah Schmiedel]

Thank you for your help, things make more sense after this explanation and when looking in the main window (ie, slice 201 of 512). 

Thanks again, 

Sarah