API Reference

The toolbox map. pycemrg-image-analysis is a stateless library: every module below accepts in-memory data and returns in-memory data — none of it reads or writes files (except the I/O utilities, which your orchestrator calls explicitly). For the design rationale and the orchestration pattern that ties these together, start with Architecture.

Core: building anatomy

Module	What it does	Page
Recipes	Named, ordered workflows (get_recipe, list_recipes) and the Recipe/WorkflowStep contracts	Recipes & Workflows
Components	Connected-component cleanup of segmentation artefacts — spatial ops + semantic wrappers	Components

Utilities: I/O & spatial queries

Module	What it does	Page
I/O	Read/write images, INR ↔ NIfTI conversion, round-trip behaviour	I/O Utilities
Spatial	Voxel ↔ physical coordinate mapping, slice extraction, mesh sampling	Spatial Queries
Label tools	LabelDiagnostic, LabelRemapper, check_labels — reconcile non-standard label schemes	Label Tools

Dataset prep: super-resolution training

Module	What it does	Page
Augmentation	Intensity & spatial augmentation for SR training pipelines	Augmentation
Metrics	Volume-comparison metrics (MSE, PSNR, SSIM, gradient error)	Metrics

Looking for the orchestration pattern?

The high-level scaffold → load → run → save sequence, the logic engines (MyocardiumLogic, ValveLogic, …), and the frozen *CreationContract dataclasses are documented on the Architecture page.