SupervoxelFeatureExtractor

Note

This page is a reference documentation. It only explains the class signature, and not how to use it. Please refer to the Habitat Guide and Python API guide (v2.0) for usage.

class SupervoxelFeatureExtractor(*args, **kwargs)

Bases: Protocol

Describe each supervoxel of one subject, replacing the default summary.

Structurally the twin of HabitatFeatureExtractor: both answer “given a partition of this subject plus the original images, what describes each region?”. They differ only in granularity and in who consumes the answer – supervoxel features feed the cohort-level clustering that DEFINES habitats, habitat features feed outcome modelling.

Separating this from Supervoxelizer is what keeps two independent scientific choices independent. It is also what makes per-supervoxel radiomics expressible at all: those families need the subject’s original intensities, which a VoxelFeatureField deliberately does not carry.

property spec: Spec

Return the algorithm specification.

__call__(subject: Subject, partition: Supervoxelization) → Supervoxelization

Recompute one subject’s per-supervoxel features.

Parameters:

• subject – Subject supplying original images when the family needs intensity information.

• partition – The subject’s supervoxel partition, whose label_array defines the regions to describe.

Returns:

The SAME partition carrying the newly computed features. Returning a Supervoxelization rather than a bare table keeps the downstream contract single-typed: the fitter and the assigner consume partitions, and never need to know which extractor described them.

__init__(*args, **kwargs)