feat: Implement no isotropic annotation

darwin/cli.py

@@ -167,6 +167,7 @@ def _run(args: Namespace, parser: ArgumentParser) -> None:
                 args.import_annotators,
                 args.import_reviewers,
                 args.overwrite,
+                isotropic=args.isotropic,
                 cpu_limit=args.cpu_limit,
             )
         elif args.action == "convert":

darwin/cli_functions.py

@@ -5,6 +5,7 @@
 import os
 import sys
 import traceback
+from functools import partial
 from glob import glob
 from itertools import tee
 from pathlib import Path
@@ -852,6 +853,7 @@ def dataset_import(
     import_annotators: bool = False,
     import_reviewers: bool = False,
     overwrite: bool = False,
+    isotropic: bool = False,
     use_multi_cpu: bool = False,
     cpu_limit: Optional[int] = None,
 ) -> None:
@@ -885,6 +887,9 @@ def dataset_import(
     overwrite : bool, default: False
         If ``True`` it will bypass a warning that the import will overwrite the current annotations if any are present.
         If ``False`` this warning will be skipped and the import will overwrite the current annotations without warning.
+    isotropic : bool, default: False
+        If ``True`` it will not resize the annotations to be isotropic.
+        If ``False`` it will resize the annotations to be isotropic.
     use_multi_cpu : bool, default: False
         If ``True`` it will use all multiple CPUs to speed up the import process.
     cpu_limit : Optional[int], default: Core count - 2
@@ -895,6 +900,10 @@ def dataset_import(
 
     try:
         importer: ImportParser = get_importer(format)
+
+        if format == "nifti" and isotropic:
+            importer = partial(importer, isotropic=True)
+
         dataset: RemoteDataset = client.get_remote_dataset(
             dataset_identifier=dataset_slug
         )

darwin/importer/formats/nifti.py

@@ -29,7 +29,9 @@
 from darwin.importer.formats.nifti_schemas import nifti_import_schema
 
 
-def parse_path(path: Path) -> Optional[List[dt.AnnotationFile]]:
+def parse_path(
+    path: Path, isotropic: bool = False
+) -> Optional[List[dt.AnnotationFile]]:
     """
     Parses the given ``nifti`` file and returns a ``List[dt.AnnotationFile]`` with the parsed
     information.
@@ -38,6 +40,9 @@ def parse_path(path: Path) -> Optional[List[dt.AnnotationFile]]:
     ----------
     path : Path
         The ``Path`` to the ``nifti`` file.
+    isotropic : bool, default: False
+        If ``True``, the function will not attempt to resize the annotations to isotropic pixel dimensions.
+        If ``False``, the function will resize the annotations to isotropic pixel dimensions.
 
     Returns
     -------
@@ -52,6 +57,11 @@ def parse_path(path: Path) -> Optional[List[dt.AnnotationFile]]:
             "Skipping file: {} (not a json file)".format(path), style="bold yellow"
         )
         return None
+    if isotropic:
+        console.print(
+            "Isotropic flag is set to True. Annotations will be resized to isotropic pixel dimensions.",
+            style="bold blue",
+        )
     data = attempt_decode(path)
     try:
         validate(data, schema=nifti_import_schema)
@@ -79,6 +89,7 @@ def parse_path(path: Path) -> Optional[List[dt.AnnotationFile]]:
             mode=nifti_annotation.get("mode", "image"),
             slot_names=nifti_annotation.get("slot_names", []),
             is_mpr=nifti_annotation.get("is_mpr", False),
+            isotropic=isotropic,
         )
         annotation_files.append(annotation_file)
     return annotation_files
@@ -92,6 +103,7 @@ def _parse_nifti(
     mode: str,
     slot_names: List[str],
     is_mpr: bool,
+    isotropic: bool = False,
 ) -> dt.AnnotationFile:
     img, pixdims = process_nifti(nib.load(nifti_path))
 
@@ -135,6 +147,7 @@ def _parse_nifti(
             processed_class_map,
             slot_names,
             pixdims=pixdims,
+            isotropic=isotropic,
         )
     if mode in ["video", "instances"]:
         annotation_classes = {
@@ -203,6 +216,7 @@ def get_mask_video_annotations(
     processed_class_map: Dict,
     slot_names: List[str],
     pixdims: Tuple[int, int, int] = (1, 1, 1),
+    isotropic: bool = False,
 ) -> Optional[List[dt.VideoAnnotation]]:
     """
     The function takes a volume and a class map and returns a list of video annotations
@@ -212,7 +226,7 @@ def get_mask_video_annotations(
     Assumptions:
     - Importing annotation from Axial view only (view_idx=2)
     """
-    new_size = get_new_axial_size(volume, pixdims)
+    new_size = get_new_axial_size(volume, pixdims, isotropic=isotropic)
 
     frame_annotations = OrderedDict()
     all_mask_annotations = defaultdict(lambda: OrderedDict())
@@ -249,9 +263,12 @@ def get_mask_video_annotations(
         slice_mask = volume[:, :, i].astype(
             np.uint8
         )  # Product requirement: We only support 255 classes!
-        slice_mask = zoom(
-            slice_mask, (new_size[0] / volume.shape[0], new_size[1] / volume.shape[1])
-        )
+
+        if isotropic:
+            slice_mask = zoom(
+                slice_mask,
+                (new_size[0] / volume.shape[0], new_size[1] / volume.shape[1]),
+            )
 
         # We need to convert from nifti_idx to raster_idx
         slice_mask = np.vectorize(
@@ -522,18 +539,25 @@ def convert_to_dense_rle(raster: np.ndarray) -> List[int]:
 
 
 def get_new_axial_size(
-    volume: np.ndarray, pixdims: Tuple[int, int, int]
+    volume: np.ndarray, pixdims: Tuple[int, int, int], isotropic: bool = False
 ) -> Tuple[int, int]:
     """Get the new size of the Axial plane after resizing to isotropic pixel dimensions.
 
     Args:
         volume: Input volume.
         pixdims: The pixel dimensions / spacings of the volume.
+        no_isotropic: bool, default: True
+            If True, the function will not attempt to resize the annotations to isotropic pixel dimensions.
+            If False, the function will resize the annotations to isotropic pixel dimensions.
 
     Returns:
         Tuple[int, int]: The new size of the Axial plane.
     """
     original_size = volume.shape
+
+    if not isotropic:
+        return original_size[0], original_size[1]
+
     original_spacing = pixdims
     min_spacing = min(pixdims[0], pixdims[1])
     return (

darwin/options.py

@@ -338,6 +338,11 @@ def __init__(self) -> None:
             action="store_true",
             help="Bypass warnings about overwiting existing annotations.",
         )
+        parser_import.add_argument(
+            "--isotropic",
+            action="store_true",
+            help="Importing annotation files with isotropic transformation.",
+        )
 
         # Cpu limit for multiprocessing tasks
         def cpu_default_types(input: Any) -> Optional[int]:  # type: ignore

tests/darwin/importer/formats/import_nifti_test.py

@@ -153,7 +153,7 @@ def test_image_annotation_nifti_import_incorrect_number_slot(
                 parse_path(path=upload_json)
 
 
-def test_image_annotation_nifti_import_single_slot_to_mask(
+def test_image_annotation_nifti_import_single_slot_to_mask_isotropic(
     team_slug_darwin_json_v2: str,
 ):
     with tempfile.TemporaryDirectory() as tmpdir:
@@ -188,7 +188,7 @@ def test_image_annotation_nifti_import_single_slot_to_mask(
             with patch("darwin.importer.formats.nifti.zoom") as mock_zoom:
                 mock_zoom.side_effect = ndimage.zoom
 
-                annotation_files = parse_path(path=upload_json)
+                annotation_files = parse_path(path=upload_json, isotropic=True)
                 annotation_file = annotation_files[0]
                 output_json_string = json.loads(
                     serialise_annotation_file(annotation_file, as_dict=False)
@@ -228,6 +228,13 @@ def test_get_new_axial_size():
     volume = np.zeros((10, 10, 10))
     pixdims = (1, 0.5, 0.5)
     new_size = get_new_axial_size(volume, pixdims)
+    assert new_size == (10, 10)
+
+
+def test_get_new_axial_size_with_isotropic():
+    volume = np.zeros((10, 10, 10))
+    pixdims = (1, 0.5, 0.5)
+    new_size = get_new_axial_size(volume, pixdims, isotropic=True)
     assert new_size == (20, 10)