[agourakis82]

Add the corrected DCMTK RGB spatial-metadata baseline for InsightSoftwareConsortium/ITK#6918. The retained Y direction matches the DICOM input; pixels, origin, spacing and size are unchanged.

Validation

Generated with the source-built ITKIODCMTKTestDriver and the existing visible-male-rgb-slice.dcm input. The corrected reader passes against this baseline and fails against the old baseline because their physical directions differ.

A separate C++23 witness checks the direction against DCMTK's DICOM metadata and verifies pixel equality with both the input and the old baseline.

SHA256: 5cd63bc548b5ff2cb1242a34a1e432c8e577c6b785c772d6bcddbf76a9c26819.

CID: bafybeiayndq33xzj6fbcr5lsb32w4s7jjomj7wsw4rvvu2zckp2ca5gsz4 (packed with ipfs-car pack --no-wrap).

The ITK content-link update will wait until this artifact can be retrieved from the official mirror.

[agourakis82]

The generic-reader correction is prepared locally without the DCMTK-specific exception. All six reader tests, plus itkDCMTKImageIOTest3 and itkDCMTKImageIOTest4, pass.

The regenerated RGB baseline matches the DICOM direction while preserving pixels, origin, spacing, and size. The old baseline fails the comparison; the corrected one passes.

The baseline is submitted in ITKTestingData#83 (https://github.com/InsightSoftwareConsortium/ITKTestingData/pull/83). I’ll update this branch once the official mirror serves the new CID.

[agourakis82]

Could a maintainer review and merge this baseline addition for official mirror ingestion? As of 2026-10-12, the official CID URL still returns HTTP 404:

https://insightsoftwareconsortium.github.io/ITKTestingData/CID/bafybeiayndq33xzj6fbcr5lsb32w4s7jjomj7wsw4rvvu2zckp2ca5gsz4

The proposed file at PR83 head 7503e4d45125f76b2e161f7066f8026fd1213678 matches SHA256 5cd63bc548b5ff2cb1242a34a1e432c8e577c6b785c772d6bcddbf76a9c26819. The prepared ITK #6918 update remains on hold until the official mirror serves those exact bytes.

[dzenanz]

Given changes in the accompanying fix PR, should this still be merged?

Also, this file is a little big. Would it benefit from some DICOM compression? And would compression be problematic for the test?

[agourakis82]

Thanks. This baseline is no longer required by the current version of ITK #6918: Identity remains the default, and the existing DCMTK baselines are retained. It should therefore not be merged as a dependency of that PR.

I have not tested DICOM compression for this file, so I cannot confirm either the size reduction or its effect on the test. Leaving this PR open for now.