prior2024_retest_perturbation

Note

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

prior2024_retest_perturbation(*, shift_fraction: float = 0.5, angle_degrees: float = 0.5, interpolator: str = 'bspline', single_resample: bool = False, warp_masks: bool = False) → PerturbationChain[source]

Simulated-retest chain of Prior et al. 2024 / MIRP 1.2.0 Appendix S2.

Paper: Prior O, et al. Identification of Precise 3D CT Radiomics for Habitat Computation by Machine Learning in Cancer. Radiol Artif Intell. 2024;6(2):e230118. doi:10.1148/ryai.230118

Order: Gaussian noise (Chang wavelet sigma) → sub-voxel translation (fraction η of voxel spacing, default 0.5, random axis signs) → 0.5° in-plane (z) rotation. Images use B-spline. single_resample=True composes translation+rotation (MIRP ≥ 2); the paper used two geometric resamples.

Default warp_masks=False matches their GitHub extractor: the perturbed CT is paired with the original ROI (compute_features_parallel_perturbed.py). Set warp_masks=True for the Zwanenburg / MIRP “image and mask move together” variant.

ROI morphological grow/shrink (MIRP perturbation_roi_adapt_size) is not in this protocol. MONAI elastic / B-spline free-form warps are a separate optional component (BSplineDeformPerturbation), not part of this chain.

Parameters:

  • shift_fraction – MIRP translation fraction in [0, 1].

  • angle_degrees – In-plane rotation in degrees.

  • interpolator – Intensity interpolator.

  • single_resample – Compose translation and rotation into one affine.

  • warp_masks – When True, nearest-neighbour warp every ROI with the image. Paper / GitHub default is False.

Returns:

A PerturbationChain.