Skip to main content

segmentation

The segmentation and classification label universes.

Which biomarker labels exist, which of them a given metric is meaningful for, and how each is displayed and coloured. Several distinct index spaces live here and must not be conflated — see the comments on the individual maps.

Module​

Functions​

extract_segmentation_area​

def extract_segmentation_area(    segmentation_areas: typing.Any, label: str, *, absent: float | None = 0.0,) ‑> float | None:

Look up one segmentation-mask area from a per-scan areas mapping.

Arguments

  • segmentation_areas: Per-scan {label: area} mapping, or a non-mapping sentinel when the model returned no segmentation data for the scan.
  • label: Segmentation-mask label to look up (e.g. "hard_drusen").
  • absent: Value returned when the mapping is present but lacks label (or holds None for it). Pass None to mark a missing label as unknown rather than a measured zero.

Returns The area as a float; absent (uncoerced) when the label is missing or its value is None; None when segmentation_areas is not a mapping.