Habitats and clustering

Habitat: sub-regions within the tumor ROI with similar imaging features; label maps use distinct colors per habitat.

Typical pipeline: preprocessing → habitat segmentation → feature extraction → (optional) machine learning.

Clustering strategies

Strategy	When to use
Two-Step (demo default)	Multi-subject studies; supervoxels per subject, then population habitats
One-Step	Single-tumor exploration
Direct Pooling	Pool all voxels across subjects (special cases)