Merge pull request #84 from ArcanaFramework/dicom-metadata

Cleans up the reading of DICOM metadata and returns dictionary not DICOM object

Author: tclose

.pre-commit-config.yaml

@@ -30,11 +30,6 @@ repos:
     rev: v1.11.2
     hooks:
       - id: mypy
-        args:
-          [
-            --strict,
-            --install-types,
-            --non-interactive,
-          ]
+        args: [--strict, --install-types, --non-interactive, --no-warn-unused-ignores]
         exclude: tests
-        additional_dependencies: [pytest, attrs, imageio]
+        additional_dependencies: [pytest, attrs, imageio, pydicom]

docs/source/developer/extensions.rst

@@ -86,7 +86,7 @@ the :class:`.WithSeparateHeader` mixin.
         ext = ".hdr"
 
         def load(self):
-            return dict(ln.split(":") for ln in self.contents.splitlines())
+            return dict(ln.split(":") for ln in self.raw_contents.splitlines())
 
     class MyFormatWithHeader(WithSeparateHeader, File):
         ext = ".myh"

extras/fileformats/extras/application/medical.py

@@ -1,31 +1,28 @@
 import typing as ty
 from pathlib import Path
-import pydicom
+from typing_extensions import TypeAlias
+import pydicom.tag
 from fileformats.core import FileSet, extra_implementation
 from fileformats.application import Dicom
 import medimages4tests.dummy.dicom.mri.t1w.siemens.skyra.syngo_d13c
 from fileformats.core import SampleFileGenerator
 
+TagListType: TypeAlias = ty.Union[
+    ty.List[int],
+    ty.List[str],
+    ty.List[ty.Tuple[int, int]],
+    ty.List[pydicom.tag.BaseTag],
+]
+
 
 @extra_implementation(FileSet.read_metadata)
 def dicom_read_metadata(
     dicom: Dicom,
-    specific_tags: ty.Optional[ty.Collection[str]] = None,
+    metadata_keys: ty.Optional[TagListType] = None,
     **kwargs: ty.Any,
 ) -> ty.Mapping[str, ty.Any]:
-    dcm = pydicom.dcmread(
-        dicom.fspath,
-        specific_tags=list(specific_tags if specific_tags is not None else []),
-    )
-    [getattr(dcm, a, None) for a in dir(dcm)]  # Ensure all keywords are set
-    metadata = {
-        e.keyword: e.value
-        for e in dcm.elements()
-        if isinstance(e, pydicom.DataElement)
-        and getattr(e, "keyword", False)
-        and e.keyword != "PixelData"
-    }
-    return metadata
+    dcm = pydicom.dcmread(dicom.fspath, specific_tags=metadata_keys)
+    return Dicom.pydicom_to_dict(dcm)
 
 
 @extra_implementation(FileSet.generate_sample_data)
@@ -38,3 +35,22 @@ def dicom_generate_sample_data(
             out_dir=generator.dest_dir
         ).iterdir()
     )
+
+
+@extra_implementation(FileSet.load)
+def dicom_load(
+    dicom: Dicom,
+    specific_tags: ty.Optional[TagListType] = None,
+    **kwargs: ty.Any,
+) -> pydicom.FileDataset:
+    return pydicom.dcmread(dicom.fspath, specific_tags=specific_tags)
+
+
+@extra_implementation(FileSet.save)
+def dicom_save(
+    dicom: Dicom,
+    data: pydicom.FileDataset,
+    write_like_original: bool = False,
+    **kwargs: ty.Any,
+) -> None:
+    pydicom.dcmwrite(dicom.fspath, data, write_like_original=write_like_original)

extras/fileformats/extras/application/serialization.py

@@ -4,7 +4,7 @@
 import yaml
 import pydra.mark
 import pydra.engine.specs
-from fileformats.core import converter, extra_implementation
+from fileformats.core import FileSet, converter, extra_implementation
 from fileformats.application import TextSerialization, Json, Yaml
 from fileformats.application.serialization import SerializationType
 
@@ -27,14 +27,18 @@ def convert_data_serialization(
     return output_format.new(output_path, dct)
 
 
-@extra_implementation(TextSerialization.load)
-def yaml_load(yml: Yaml) -> SerializationType:
+@extra_implementation(FileSet.load)
+def yaml_load(yml: Yaml, **kwargs: ty.Any) -> SerializationType:
     with open(yml.fspath) as f:
         data = yaml.load(f, Loader=yaml.Loader)
     return data  # type: ignore[no-any-return]
 
 
-@extra_implementation(TextSerialization.save)
-def yaml_save(yml: Yaml, data: SerializationType) -> None:
+@extra_implementation(FileSet.save)
+def yaml_save(
+    yml: Yaml,
+    data: SerializationType,
+    **kwargs: ty.Any,
+) -> None:
     with open(yml.fspath, "w") as f:
-        yaml.dump(data, f)
+        yaml.dump(data, f, **kwargs)

extras/fileformats/extras/application/tests/test_application_medical.py

@@ -1,8 +1,18 @@
+import pytest
 from fileformats.application import Dicom
 
 
 def test_dicom_metadata():
 
     dicom = Dicom.sample()
 
-    assert dicom.metadata["EchoTime"] == "2.07"
+    assert dicom.metadata["EchoTime"] == 2.07
+
+
+def test_dicom_metadata_with_specific_tags():
+
+    dicom = Dicom(Dicom.sample(), metadata_keys=["EchoTime"])
+
+    assert dicom.metadata["EchoTime"] == 2.07
+    with pytest.raises(KeyError):
+        dicom.metadata["PatientName"]

extras/fileformats/extras/application/tests/test_application_serialization.py

@@ -3,15 +3,10 @@
 from fileformats.application import Json, Yaml
 
 
-SAMPLE_JSON = """{
-    "a": "string field",
-    "alist": [0, 1, 2, 3, 4, 5],
-    "anesteddict": {
-        "x": null,
-        "y": [],
-        "z": 42.0
-    }
-}"""
+SAMPLE_JSON = (
+    """{"a": "string field", "alist": [0, 1, 2, 3, 4, 5], """
+    """"anesteddict": {"x": null, "y": [], "z": 42.0}}"""
+)
 
 SAMPLE_YAML = """a: string field
 alist:
@@ -38,7 +33,7 @@ def test_json_to_yaml(work_dir):
         f.write(SAMPLE_JSON)
     jsn = Json(in_file)
     yml = Yaml.convert(jsn)
-    assert yml.contents == SAMPLE_YAML
+    assert yml.raw_contents == SAMPLE_YAML
 
 
 # @pytest.mark.xfail(
@@ -50,5 +45,5 @@ def test_yaml_to_json(work_dir):
     with open(in_file, "w") as f:
         f.write(SAMPLE_JSON)
     yml = Yaml(in_file)
-    Json.convert(yml)
-    assert yml.contents == SAMPLE_JSON
+    jsn = Json.convert(yml)
+    assert jsn.raw_contents == SAMPLE_JSON

extras/fileformats/extras/image/readwrite.py

@@ -1,15 +1,18 @@
 import imageio
+import typing as ty
 import numpy  # noqa: F401
 import typing  # noqa: F401
-from fileformats.core import extra_implementation
+from fileformats.core import FileSet, extra_implementation
 from fileformats.image.raster import RasterImage, DataArrayType
 
 
-@extra_implementation(RasterImage.load)
-def read_raster_data(image: RasterImage) -> DataArrayType:
+@extra_implementation(FileSet.load)
+def read_raster_data(image: RasterImage, **kwargs: ty.Any) -> DataArrayType:
     return imageio.imread(image.fspath)  # type: ignore
 
 
-@extra_implementation(RasterImage.save)
-def write_raster_data(image: RasterImage, data: DataArrayType) -> None:
-    imageio.imwrite(image.fspath, data)
+@extra_implementation(FileSet.save)
+def write_raster_data(
+    image: RasterImage, data: DataArrayType, **kwargs: ty.Any
+) -> None:
+    imageio.imwrite(image.fspath, data, **kwargs)

fileformats/application/medical.py

@@ -1,6 +1,10 @@
+import typing as ty
 from fileformats.generic import BinaryFile
 from fileformats.core.mixin import WithMagicNumber
 
+if ty.TYPE_CHECKING:
+    import pydicom
+
 
 class Dicom(WithMagicNumber, BinaryFile):
 
@@ -10,3 +14,57 @@ class Dicom(WithMagicNumber, BinaryFile):
     binary = True
 
     alternate_exts = (".dcm",)  # dcm is recommended not required
+
+    @classmethod
+    def pydicom_to_dict(
+        cls, dcm: "pydicom.Dataset", omit: ty.Collection[str] = ("PixelData",)
+    ) -> ty.Dict[str, ty.Any]:
+        """Convert a pydicom Dataset to a dictionary.
+
+        Parameters
+        ----------
+        dcm : pydicom.Dataset
+            The pydicom Dataset to convert.
+        omit : Collection[str], optional
+            A collection of keys to omit from the dictionary, by default ("PixelData",)
+
+        Returns
+        -------
+        Dict[str, Any]
+            The dictionary representation of the pydicom Dataset
+        """
+        import pydicom.dataset
+        import pydicom.valuerep
+        import pydicom.multival
+        import pydicom.uid
+
+        # Ensure that all keys are loaded before creating dictionary otherwise the keywords
+        # will not be set in the elem
+        [getattr(dcm, attr, None) for attr in dir(dcm)]
+        dct: ty.Dict[str, ty.Any] = {}
+        for elem in dcm.values():
+            try:
+                key = elem.keyword  # type: ignore[union-attr, attr-defined]
+            except AttributeError:
+                key = None
+            if not key:
+                key = elem.tag.json_key  # type: ignore[attr-defined]
+            if key not in omit:
+                value = elem.value  # type: ignore[attr-defined]
+                if isinstance(value, pydicom.multival.MultiValue):
+                    value = [str(v) for v in value]
+                elif isinstance(value, pydicom.uid.UID):
+                    value = str(value)
+                elif isinstance(value, bytes):
+                    value = value.decode(errors="ignore")
+                elif isinstance(value, pydicom.dataset.Dataset):
+                    value = cls.pydicom_to_dict(value, omit)
+                elif isinstance(value, pydicom.valuerep.IS):
+                    value = int(value)
+                elif isinstance(value, pydicom.valuerep.DSfloat):
+                    value = float(value)
+                # Can be handy to be able to access family_name and given_name separately
+                # elif isinstance(value, pydicom.valuerep.PersonName):
+                #     value = str(value)
+                dct[key] = value
+        return dct

fileformats/application/serialization.py

@@ -103,16 +103,16 @@ def generate_yaml_sample_data(
 
 
 @extra_implementation(FileSet.load)
-def load(jsn: Json) -> SerializationType:
+def load(jsn: Json, **kwargs: ty.Any) -> SerializationType:
     try:
         with jsn.open() as f:
-            dct: ty.Dict[str, ty.Any] = json.load(f)
+            dct: ty.Dict[str, ty.Any] = json.load(f, **kwargs)
     except json.JSONDecodeError as e:
         raise FormatMismatchError(f"'{jsn.fspath}' is not a valid JSON file") from e
     return dct
 
 
 @extra_implementation(FileSet.save)
-def save(jsn: Json, data: SerializationType) -> None:
+def save(jsn: Json, data: SerializationType, **kwargs: ty.Any) -> None:
     with jsn.open("w") as f:
-        json.dump(data, f)
+        json.dump(data, f, **kwargs)