* fix(pt): resolve Philips PET private SUV bulkdata before scaling
When a DICOMweb server delivers the Philips PET private tags SUVScaleFactor
(7053,1000) / ActivityConcentrationScaleFactor (7053,1009) as bulkdata, dcmjs
naturalization leaves them as { BulkDataURI } objects. These were fed verbatim
to calculate-suv, which treats the object as a valid value and silently
corrupts the SUV scaling factors.
Resolve these scalar private tags to numbers during ingestion - in both the
lazy (async) and non-lazy (sync) DICOMweb metadata paths, before INSTANCES_ADDED
fires - by decoding the bulkdata (VR-aware: DS/IS text or little-endian FL/FD).
Harden getPTImageIdInstanceMetadata to coerce values to finite numbers (reusing
@ohif/core utils.toNumber) and reject unresolved bulkdata objects so they can
never reach calculate-suv. Share the bulkdata-attach helper between both
metadata paths. Adds unit tests for the bulkdata decoder/resolver and for
getPTImageIdInstanceMetadata.
* refactor(bulkdata): move PET bulkdata resolution to a generic core tag registry
Generalize resolvePETPrivateScalarBulkData into a datasource-agnostic
utils.resolveBulkDataTags in @ohif/core, backed by a static tag registry
seeded with the Philips PET SUV/activity-concentration scalar tags and
extensible via registerResolvedBulkDataTags.
* fix(bulkdata): strip NUL padding and refresh qido auth before resolution
Address review feedback:
- decodeText now strips NUL (0x00) padding, which String.trim() leaves
intact, so NUL-padded DS/IS values no longer decode as NaN. Adds a
regression test.
- refresh qidoDicomWebClient.headers before resolveBulkDataTags in both
series-metadata paths; retrieveBulkData is bound to qidoDicomWebClient,
matching every other qido op in this file.
* fix(dicomweb): await deferred metadata storage
* fix(dicomweb): support single-part bulkdata responses
70421225d7