def _download_treatmentrecord_test_file():
data_paths = download.zip_data_paths("treatmentrecord-anonymisation.zip")
return data_paths
def get_dataset_metadata():
release_url = "https://github.com/pymedphys/data/releases/download/structure-dicom"
dicom_zip_url_pattern = f"{release_url}/" + "{dicom_type}.{uid}_Anonymised.zip"
mappings_url = f"{release_url}/mappings.zip"
data_download_root = pathlib.Path("auto-segmentation-dicom")
save_filename = data_download_root.joinpath(get_filename_from_url(mappings_url))
mappings_paths = download.zip_data_paths(
save_filename,
check_hash=True,
redownload_on_hash_mismatch=True,
url=mappings_url,
)
mapping_filenames = {
"hashes": "hashes.json",
"name_aliases": "name_mappings.json",
"paths_and_uids": "uid-cache.json",
"structure_names_by_uid": "structure-names-mapping-cache.json",
}
mapping_paths = {}
for label, filename in mapping_filenames.items():
mapping_path = [path for path in mappings_paths if path.name == filename]
if len(mapping_path) != 1:
raise ValueError(f"Expected exactly one file named {filename}.")
mapping_paths[label] = mapping_path[0]
hash_path = mapping_paths["hashes"]
data_path_root = hash_path.parent.parent
with open(mapping_paths["paths_and_uids"]) as f:
uid_cache = json.load(f)
(
ct_image_paths,
structure_set_paths,
ct_uid_to_structure_uid,
structure_uid_to_ct_uids,
) = indexing.read_uid_cache(data_path_root, uid_cache)
with open(mapping_paths["structure_names_by_uid"]) as f:
structure_names_cache = json.load(f)
structure_names_by_ct_uid = structure_names_cache["structure_names_by_ct_uid"]
structure_names_by_structure_set_uid = structure_names_cache[
"structure_names_by_structure_set_uid"
]
uid_to_url = {}
for structure_uid, ct_uids in structure_uid_to_ct_uids.items():
uid_to_url[structure_uid] = dicom_zip_url_pattern.format(
dicom_type="RS", uid=structure_uid
)
for ct_uid in ct_uids:
uid_to_url[ct_uid] = dicom_zip_url_pattern.format(
dicom_type="CT", uid=ct_uid
)
names_map = filtering.load_names_mapping(mapping_paths["name_aliases"])
return (
data_path_root,
structure_set_paths,
ct_image_paths,
ct_uid_to_structure_uid,
structure_uid_to_ct_uids,
names_map,
structure_names_by_ct_uid,
structure_names_by_structure_set_uid,
uid_to_url,
hash_path,
)
def _download_rtplan_test_file():
data_paths = download.zip_data_paths("rtplan-anonymisation.zip")
return data_paths