v0.1.0

Initial public release. Establishes the core anatomy pipeline, label management
tooling, ML-oriented utilities, and supporting infrastructure.

Anatomy Pipeline

• MyocardiumLogic — derives myocardium masks from blood pool labels using
  physical-space distance maps and configurable MyocardiumRule contracts.
  Supports multi-step application sequences (ApplicationStep) for fine-grained
  control over how walls are written into the output image.
• ValveLogic — creates valve masks via signed-distance intersection between two
  anatomical structures, driven by ValveRule contracts.
• RingLogic — creates pulmonary vein ring labels by combining vein distance maps
  with atrium myocardium boundaries, using RingRule contracts.
• SegmentationLogic — creates cylinder masks from point clouds; used for
  fibre-direction annotation and cutting plane generation.
• schematics/ — pre-defined anatomy blueprints for LV, RV, LA, RA, valves, and
  vein rings. Used as reference configurations for contract construction and
  label diagnostics.
• recipes.py — six named end-to-end workflows: biventricular_basic,
  biventricular_full, four_chamber_basic, four_chamber_full, and variants.
  Each recipe is a pre-sequenced list of logic calls with fixed defaults.
• logic/contracts.py — all data contracts as frozen dataclasses:
  MyocardiumCreationContract, ValveCreationContract, RingCreationContract,
  CylinderCreationContract, ApplicationStep, MyocardiumRule, ValveRule,
  RingRule.
• logic/builders.py — SegmentationPathBuilder constructs contracts from
  file paths and image metadata; reduces boilerplate in orchestrators.

Label Management

• utilities/label_tools.py — LabelDiagnostic compares an image's present
  integer labels against a named schematic and produces a DiagnosticReport
  listing missing and unexpected labels. LabelRemapper converts a
  DiagnosticReport into an integer-to-integer mapping via
  suggest_mapping_from_report().
• scaffolding.py — ConfigScaffolder generates YAML and JSON parameter files
  for workflow runs. scaffold_components_with_mapping() handles the common case
  where 3D Slicer has reset label values to sequential 1–N.
• scripts/utilities/labels_cli.py — command-line tool for inspecting label
  contents and computing volume statistics from NIfTI segmentation files.

Utilities

• utilities/masks.py — four mask composition modes (ADD, REPLACE,
  REPLACE_ONLY, REPLACE_EXCEPT), wired through dispatchers.py.
• utilities/components.py — keep_largest_component (per-label cleanup) and
  keep_largest_structure (multi-label debris removal).
• utilities/spatial.py — point sampling on image grids, bounding-box slicing,
  coordinate-system utilities.
• utilities/geometry.py — cylinder mask calculation in physical space.
• utilities/filters.py — morphological and intensity filtering helpers.
• utilities/io.py — load_image / save_image wrappers; consistent spacing
  and metadata handling across SimpleITK and NumPy array forms.
• utilities/postprocessing.py — label volume computation and related cleanup.
• scripts/utilities/inr_to_nifti.py — converts legacy .inr format images to
  NIfTI, preserving physical spacing metadata.

ML Utilities

• utilities/metrics.py — Dice coefficient, SSIM, and related volume comparison
  metrics. All functions expect input normalized to [0, 1] in (Z, Y, X) order.
• utilities/augmentation.py — spatial and intensity augmentation (flips, rotations,
  noise, contrast perturbation). Validates [0, 1] normalization at entry.
• utilities/artifact_simulation.py — downsample_volume simulates
  lower-resolution acquisitions; preserve_extent flag controls whether physical
  extent is held fixed (needed for ML coordinate normalization pipelines).
• utilities/sampling.py — extract_center_patch with automatic padding for
  patch-based training pipelines.

Infrastructure

• PyPI-ready packaging via pyproject.toml + setuptools. Package name:
  pycemrg-image-analysis, requires Python >= 3.10.
• GitHub Actions workflow for automated PyPI publishing on tag push.
• MkDocs-based API documentation with per-module reference pages and guides.
• Unit test suite covering components, metrics, augmentation, artifact simulation,
  sampling, spatial coordinates, and scaffolding — no test data required.
• Integration test suite covering myocardium, valve, and ring creation — requires
  PYCEMRG_TEST_DATA_ROOT to be set; skipped automatically when absent.

Dependencies

Requires pycemrg>=0.1.0 (separate repository) for LabelManager and
ConfigScaffolder. Key third-party dependencies: SimpleITK, NumPy, SciPy,
scikit-image, trimesh, meshio, nibabel, pydicom.