CT/lung/src/preprocessing/build_lidc_masks.py

from __future__ import annotations

import argparse
from pathlib import Path

import numpy as np


def average_expert_masks(mask_paths: list[str | Path], output_path: str | Path) -> Path:
    """Average binary expert masks into the soft supervision mask used by Stage A."""

    if not mask_paths:
        raise ValueError("At least one expert mask is required")
    arrays = []
    for path in mask_paths:
        data = np.load(path)
        arrays.append(np.asarray(data["mask"] if "mask" in data else data[data.files[0]], dtype=np.float32))
    shape = arrays[0].shape
    if any(arr.shape != shape for arr in arrays):
        raise ValueError("All expert masks must have the same shape before averaging")
    soft = np.mean(np.stack(arrays, axis=0), axis=0)
    output_path = Path(output_path)
    output_path.parent.mkdir(parents=True, exist_ok=True)
    np.savez_compressed(output_path, mask=soft.astype(np.float32))
    return output_path


def main() -> None:
    parser = argparse.ArgumentParser(description="Average LIDC expert mask NPZ files.")
    parser.add_argument("--masks", nargs="+", required=True)
    parser.add_argument("--out", required=True)
    args = parser.parse_args()
    average_expert_masks(args.masks, args.out)


if __name__ == "__main__":
    main()