Release notes

0.5.3

  • Official FHIR terminology: DICOM codes are resolved from the pinned official fhir.dicom package, while all 73 ADA 1100 image types are resolved from the CodeSystems at terminology.open-ortho.org.

  • Offline terminology lock: source IDs and versions are recorded in the committed dicom4ortho/_generated_codes.py module. Normal builds and runtime use do not access the network.

  • Retired terminology CSV files: codes.csv and image_types.csv are no longer packaged. views.csv remains the local view-layout source.

  • Correct CID 4070 events: treatment progress now uses PatientRegistration (184047000), OrthodonticTreatmentStarted (1332161000), and OrthodonticTreatmentStopped (1340210007).

  • Date-based treatment progress: set_treatment_progress_from_date(event_type, event_date) calculates the TID 3465 offset from the photograph acquisition date. Constructor metadata accepts treatment_event_date with acquisition_datetime.

  • Acquisition-time precedence: an explicit acquisition_datetime overrides EXIF. If it is omitted, EXIF DateTimeOriginal is loaded before treatment progress is calculated and used as the fallback acquisition time.

  • Compatibility: the days-based API and the old OrthodonticTreatment and Posttreatment event names remain available with DeprecationWarning until version 2.

  • Resource maintenance: make update_resources explicitly fetches FHIR terminology and regenerates the lock without committing it. Updating views.csv remains a separate make fetch_resources operation.

  • Practice personnel: dental_provider_firstname and dental_provider_lastname now encode the orthodontist or dentist responsible for treatment as DICOM Physicians of Record. Explicit operator metadata encodes the clinical staff member as Operators’ Name, while an actual referring physician is copied from MWL without conflating it with the requesting physician.

  • MWL conversion: the controller’s image-plus-worklist conversion now copies valid worklist attributes and returns the generated photograph.

  • JPEG 2000 preserved as received: the source codestream is encapsulated unchanged instead of being decoded and re-encoded, and the Transfer Syntax and LossyImageCompression follow its wavelet transform. A JP2 container is unwrapped to its jp2c codestream first.

0.5.2

  • CP-1570 / TID 3465 compliance: OrthodonticPhotograph now builds AcquisitionContextSequence from typed OrthoView fields mapping directly to TID 3465 rows (OrthognathicFunctionalCondition, FindingByInspection, ObservableEntity, DentalOcclusion, and optional LongitudinalTemporalEventType + Offset via set_treatment_progress()).

  • Build-time code generation: replaced runtime CSV parsing with a code-generation pipeline (tools/generate_codes.py). Reads codes.csv and views.csv and emits dicom4ortho/_generated_codes.py — a committed lock file of typed Python constants (DicomCode, OrthoView). No network I/O at runtime.

  • Typed data layer: two new frozen dataclasses in m_dent_oip.py — DicomCode (with to_dataset() / to_sequence()) and OrthoView (all 73 view definitions with typed fields for every DICOM attribute).

  • Incomplete-view enforcement: views with a variable ViewCodeSequence (IV28, IV30, EV40) now raise TypeError immediately at construction unless view_code_keyword is supplied. set_view_code() fills the slot and auto-attaches any predefined ViewModifierCodeSequence.

  • codes.csv / views.csv corrections: OrthognathicFunctionalConditions concept-name code corrected (ZZZZZZ → 130325, DCM, per CP-1570); closeup keyword mismatch fixed; IV29 given projection_frontal as its fixed ViewCode.

  • CLI fix: dicom4ortho list-image-types now works correctly (fixed TypeError: 'PosixPath' object is not iterable in print_image_types()); help text updated to show the full command.

  • CLI cleanup: removed --url-codes / --url-views flags (broken since the refactor; resource updates are now a build-time operation via make update_resources).

  • Dev environment: flake.nix rewritten to use pip/pyproject.toml as the single source of truth for Python deps; dciodvfy and dcmtk added to the Nix shell.

  • Makefile: make update_resources now only regenerates _generated_codes.py from committed CSVs; new make fetch_resources target does the upstream curl downloads separately. New make help default target.

  • Tests: 29 new tests in test/test_generated_codes.py covering CSV consistency, all 73 view attribute correctness, and the view_code_keyword / set_view_code() API.

0.5.1

  • Rename get/set_image_type_code_sequence → get/set_image_type_code_dataset for consistency with pydicom naming conventions.

  • Add tests for image type code dataset handling in OrthodonticPhotograph.

  • Refactor test files for improved organisation and logging consistency.

0.4.7

  • Fix default Series and Study Date and Time to empty.

0.4.2

  • Add partial support for EXIF metadata: copy tags from EXIF to DICOM according to standard. Only Manufacturer related and date related tags are copied.

0.4

  • Add support for DICOM MWL (Modality Worklist): copy tags from MWL

  • Add support for using InstanceNumber and ScheduledProtocolCode to define image type

  • Add method to decode the image type from the InstanceNumber and ScheduledProtocolCodeSequence

  • Remove code related to teeth

  • Add setters for Series and Instance Numbers

0.2

  • Add BurnedInAnnotation, Operator, Institution and Performing Physician tags

  • Load views and codes from ADA-1107

  • Fix Patient, Doctor and Operator names

  • Add support for JPEG2000

  • UID is now generated from hashing the input file

  • Other bug fixes