1. Summary of problem description

(Originally, the bug described only a test image that could not be found. For this old description see ChangeRequests/7210a)

mitkDICOMLocaleTest discovered flaws in reading Secondary Capture (SC) DICOM images

1.1. Actual behavior

Pixel spacing of test image is read as 1.0 where 3.14... is expected

1.2. Expected behavior

Spacing should be consistent with data found in DICOM tags.

2. Cause of the bug

DicomSeriesReader was not build with Secondary Capture images in mind and not tested on them.

As for the wrong spacing, GDCM expects Secondary Capture DICOM images to declare pixel spacing in tag (0018,2010) instead of (0028,0030). Our test image did not provide the expected tag.

3. Proposed solution

Adapt test data to make test pass. Enhance test data by new CT and MR images. Add s short statement on the SC reading limitation.

Rationale:

• - MITK's main purpose is not to make measurements on Secondary Capture images, so
  • we can live with wrong spacings
  - the test image was hand-crafted and is not a 'nice' DICOM image. We do not know
  • whether actual Secondary Capture images from other devices use the (0018,2010) or the (0028,0030) tag for storing pixel spacing
  - changing GDCM's code to test (0028,0030) as a secondary spacing hint would be
  • too much effort for too little gain

4. Affected classes

DicomSeriesReader and DicomLocaleTest

5. How will the bugfix get tested?

Continuous and nightly test execution. Manual execution during fixing the issue.