plismbench.engine.extract.utils module#

Utility functionalities for the extraction pipeline.

plismbench.engine.extract.utils.sort_coords(slide_features: ndarray) → ndarray[source]#

Sort slide features by coordinates.

plismbench.engine.extract.utils.save_features(slide_features: list[ndarray], slide_id: str, export_path: Path) → None[source]#

Save features to disk.

Parameters:

  • slide_features (list[numpy.ndarray]) – Current slide features.

  • slide_id (str) – Current slide id.

  • export_path (pathlib.Path) – Export path for slide features.

plismbench.engine.extract.utils.process_imgs(imgs: Tensor, tile_ids: list[str], model: Module) → ndarray[source]#

Perform inference on input (already transformed) images.

Parameters:

  • imgs (torch.Tensor) – Transformed images (e.g. normalized, cropped, etc.).

  • tile_ids (list[str]:) – List of tile ids.

  • model (torch.nn.Module) – Feature extractor.