cellmil.data.storage

Classes

Storage(wsi_name, output_path, metadata, ...)

Storage class to store all WSI related files

class cellmil.data.storage.Storage(wsi_name: str, output_path: Union[Path, str], metadata: dict[str, Any], mask_images: dict[str, PIL.Image.Image], mask_images_annotations: dict[str, PIL.Image.Image], thumbnails: dict[str, PIL.Image.Image], store_masks: bool = False, save_context: bool = False, context_scales: Optional[List[int]] = None)

Bases: object

Storage class to store all WSI related files

Generates the following folder structure for storage:

• Output-Path/WSI-Name

  • metadata.yaml: Metadata of the WSI

  • annotation_masks: preview images of annotations

  • patches: store extracted patches with each path “wsi_name_row_col.png”

  • metadata: store metadata for each path “wsi_name_row_col.yaml”

  • thumbnails: WSI thumbnails

  • tissue masks: Masks of tissue detection

  • Optional: context: context folder with subfolder for each context scale

  • Optional: masks: Masks for each patch as .npy files (numpy arrays)

Parameters:

• wsi_name (str) – Name of the WSI, as string. Just the name without suffix and no path!

• output_path (Union[Path, str]) – Path to the folder where the resulting dataset should be stored.

• metadata (dict) – Metadata of the WSI. Is stored in parent directory

• mask_images (dict[str, Image]) – Masks generated during tissue detection stored in dict with keys equals the mask name and values equals the PIL image

• mask_images_annotations (dict[str, Image]) – Annotation masks for provided annotations for the complete WSI. Masks are equal to the tissue masks sizes. Keys are the mask names and values are the PIL images.

• thumbnails (dict[str, Image]) – Dictionary with thumbnails and corresponding thumbnail names. Names are keys, PIL Images are values

• store_masks (bool, optional) – Set to store masks per patch. Defaults to False.

• save_context (bool, optional) – If context patches are provided. Defaults to False.

• context_scales (List[int], optional) – List with context scales. Defaults to None.

__init__(wsi_name: str, output_path: Union[Path, str], metadata: dict[str, Any], mask_images: dict[str, PIL.Image.Image], mask_images_annotations: dict[str, PIL.Image.Image], thumbnails: dict[str, PIL.Image.Image], store_masks: bool = False, save_context: bool = False, context_scales: Optional[List[int]] = None) → None

save_meta_data() → None

Store arbitrary meta data in a yaml file on wsi output folder

save_masks(mask_images: Dict[str, Image])

Save tissue masks

Parameters:

mask_images (dict[str, Image]) – Masks generated during tissue detection stored in dict with keys equals the mask name and values equals the PIL image

save_annotation_mask(mask_images_annotations: Dict[str, Image])

Save annotation masks

Parameters:

mask_images_annotations (dict[str, Image]) – Annotation masks for provided annotations for the complete WSI. Masks are equal to the tissue masks sizes. Keys are the mask names and values are the PIL images.

save_thumbnails(thumbnails: Dict[str, Image])

Save thumbnails of WSI

Parameters:

thumbnails (dict[str, Image]) – Dictionary with thumbnails and corresponding thumbnail names. Names are keys, PIL Images are values

save_elem_to_disk(patch_result: tuple[numpy.ndarray[Any, Any], dict[str, Any], None, dict[str, Any]]) → None

clean_up(patch_distribution: dict[int, int], patch_metadata_list: list[dict[str, dict[str, Any]]])

Clean-Up function, called after WSI has been processed. Appends WSI to processed.json file and generated a metadata file in root folder called patch_metadata.json with merged metadata for all patches in one file.

Parameters:

• patch_distribution (dict) – Patch distrubtion dict. Keys: Lables, values: number of patches in class

• patch_metadata_list (list[dict]) – List with all patch metadata to store