Skip to main content

scan_columns

Column names for the per-scan metrics the calculation steps emit.

The flat column names a metrics dataclass projects into a DataFrame, plus the helpers that build the per-label names dynamically (a biomarker's probability, segmentation area and consecutive-B-scan run each expand to one column per label, so they cannot be spelled out as constants).

Module​

Functions​

biomarker_area_col_name​

def biomarker_area_col_name(label: str) ‑> str:

Flat column name for a per-biomarker segmentation area (mm²).

lesion_group_col_name​

def lesion_group_col_name(group: str, metric: str) ‑> str:

Flat column name for one metric of one area group.

Arguments

  • group: The configured group name (e.g. "drusen").
  • metric: One of LESION_GROUP_METRIC_SUFFIXES.

Returns The <group>_<metric> column name.

max_pathology_prob_col_name​

def max_pathology_prob_col_name(pathology_name: str) ‑> str:

Generate the max probability column name for a given pathology.

n_scan_run_col_name​

def n_scan_run_col_name(label: str) ‑> str:

Flat column name for a biomarker's longest consecutive-B-scan run.