Type selection: short vs ushort

[KH]

Hi,

At 'Open Main Image' window, when I select 'DICOM Image Series' from the 'File Format' menu, it seems to automatically select '16 bit signed integer (short)' as the data type, resulting in an unwanted dynamic range. High pixel values are thus clipped (note the blue 'blobs' in the middle of the red area, at the cursor's cross):

However, when I select  'DICOM Single Image'  from the 'File Format' menu,  , it seems to automatically select '16 bit unsigned integer (ushort)' as the data type, as it should. Now the dynamic range is correct and no clipping of high pixel value occur (note that there are no blue 'blobs' in the middle of the red area as before, at the cursor's cross):

How can I specify the data type (16 bit unsigned integer (ushort)) for a series of images? Can it be corrected?

Thank you.

(A dcm file of the above screenshots is attached)

[Paul Yushkevich]

Hello

Does your image open correctly in other Dicom viewer software, like Horos?

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 view this discussion on the web visit https://groups.google.com/d/msgid/itksnap-users/a83fd7ac-11d5-4e66-b9ed-791750a3e9cdn%40googlegroups.com.

--

Paul A. Yushkevich, Ph.D.
Professor of Radiology
Penn Image Computing and Science Laboratory
University of Pennsylvania Perelman School of Medicine

[KH]

Hello Prof. Yushkevich,

Horos is not available for me as I use windows. This image does open correctly in python (pydicom.dcmread) and in 3D Slicer.

In ITK-SNAP it opens correctly only when selecting 'DICOM Single Image' from the 'File Format' menu, but fails when selecting  'DICOM Image Series'  (high pixel values get saturated).

When selecting  'DICOM Single Image' the data type is automatically set to be 'short', but when selecting 'DICOM Image Series' the data type is set to be 'double', thus values above 2^15-1=32767 are displayed correctly, rather than negative is in 'short'.

I attached the image to the my first post in this thread.

Thank you!

ב-יום שלישי, 16 בפברואר 2021 בשעה 15:41:26 UTC+2, pyush...@gmail.com כתב/ה:

[Jilei Hao]

Hi KH,

The Pixel Representation tag (0028 | 0103) in the sample image has value 1. From dicom standard, it means the pixel value is signed (two's complement). That's why itk-snap is loading it as signed short (int16). If you change the value from 1 to 0 for all series images, it will correctly load as unsigned short (uint16).

Related standard page:

http://dicom.nema.org/medical/Dicom/2018d/output/chtml/part03/sect_C.7.6.3.html

Best,

Jilei