[DAR-6479][External] Get primary_plane from medical metadata of the slot during nifti export

[vvihorev]

Problem

Non-axial DICOM files with extracted views and at least one mask annotation break during export in nifti format. We need to know which is the primary_plane for non-axial files in Darwin Py to export them properly. The primary_plane (not axial) determined by the volumetric extractor is not stored in annotation files.

DICOM files uploaded without exporting views have one slot. If the file is non-axial this slot would have the primary plane in the plane map in metadata e.g. plane_map: {"0": "CORONAL"}.

However, files uploaded with views exported have three slots, and always have plane_map: {"0.1": "AXIAL", "0.2": "CORONAL", "0.3": "SAGITTAL"}. Taking the plane name from the first slot is incorrect, and we need to know which plane is the primary one explicitly.

PR to support primary_plane storage in annotations on the backend

Solution

We get primary_plane from medical metadata of DICOM annotation files. We use this primary_plane set by the volumetric extractor instead of the plane of the first slot to populate volumes from raster layer during nifti export.

Changelog

Use primary_plane from medical metadata of DICOM annotation files to populate volumes from raster layer during nifti export

[linear]

DAR-6479 BUG: NIfTi exports not working on non-axial files

[Copilot]

Pull Request Overview

This PR updates the nifti export process to correctly determine the primary_plane for non-axial DICOM files by reading the value from the medical metadata instead of relying solely on the plane_map.

• Updated primary_plane retrieval to check medical metadata first.
• Retained fallback and legacy handling to ensure backward compatibility.