Traditional radiomics

Note

Whole-ROI radiomics without habitat maps. Guide: Feature extraction API. Features on habitat maps: Habitat features.

Goal: whole-ROI PyRadiomics without habitat maps. For habitat features use Feature extraction (CLI / YAML). ROI-level radiomics, voxel-level radiomics, and 3-D shape match PyRadiomics execute(); the digital-phantom table and alignment notes are on Traditional Radiomics.

Run the demo

habit check-config --config config/radiomics/config_traditional_radiomics.yaml
habit radiomics --config config/radiomics/config_traditional_radiomics.yaml

Your data

★ Edit paths.images_folder (folder with images/ + masks/), paths.out_dir, and processing.process_image_types (modality names).

Success: feature tables under paths.out_dir.

Habitat-wise tables (when you do have maps) overlay the same anatomy. The figure is not from habit radiomics above (whole-ROI has no labels). It is written by the feature-extraction gallery (Habitat features). Reproduce it:

python docs/source/examples/scripts/feature_extraction_demo.py

The plot call in that script (ROI = "LAP"):

from habit.viz import plot_habitat_overlay

fig = plot_habitat_overlay(subject.image(ROI), habitat_map, title="habitats")

Same file the gallery script writes to out/feature_extract_overlay.png.