Merge pull request #20 from ArcanaFramework/brainhack2023

Started to add in radlex classifiers and fixed up extras hooks

.gitignore

@@ -22,3 +22,4 @@ __pycache__
 ~*
 *.venv
 /fileformats/medimage/_version.py
+/extras/fileformats/extras/medimage/_version.py

extras/fileformats/extras/medimage/dicom.py

@@ -11,22 +11,22 @@
 
 
 @MedicalImage.read_array.register
-def dicom_read_array(collection: DicomCollection):
+def dicom_read_array(collection: DicomCollection) -> np.ndarray:
     image_stack = []
     for dcm_file in collection.contents:
         image_stack.append(pydicom.dcmread(dcm_file).pixel_array)
     return np.asarray(image_stack)
 
 
 @MedicalImage.vox_sizes.register
-def dicom_vox_sizes(collection: DicomCollection):
+def dicom_vox_sizes(collection: DicomCollection) -> ty.Tuple[float, float, float]:
     return tuple(
         collection.metadata["PixelSpacing"] + [collection.metadata["SliceThickness"]]
     )
 
 
 @MedicalImage.dims.register
-def dicom_dims(collection: DicomCollection):
+def dicom_dims(collection: DicomCollection) -> ty.Tuple[int, int, int]:
     return tuple(
         (
             collection.metadata["Rows"],
@@ -37,12 +37,12 @@ def dicom_dims(collection: DicomCollection):
 
 
 @DicomCollection.series_number.register
-def dicom_series_number(collection: DicomCollection):
+def dicom_series_number(collection: DicomCollection) -> str:
     return str(collection.metadata["SeriesNumber"])
 
 
 @FileSet.generate_sample_data.register
-def dicom_dir_generate_sample_data(dcmdir: DicomDir, dest_dir: Path, seed: ty.Union[int, Random], stem: ty.Optional[str]):
+def dicom_dir_generate_sample_data(dcmdir: DicomDir, dest_dir: Path, seed: ty.Union[int, Random] = 0, stem: ty.Optional[str] = None) -> ty.Iterable[Path]:
     dcm_dir = medimages4tests.dummy.dicom.mri.t1w.siemens.skyra.syngo_d13c.get_image()
     # Set series number to random value to make it different
     if isinstance(seed, Random):
@@ -60,7 +60,7 @@ def dicom_dir_generate_sample_data(dcmdir: DicomDir, dest_dir: Path, seed: ty.Un
 
 
 @FileSet.generate_sample_data.register
-def dicom_set_generate_sample_data(dcm_series: DicomSeries, dest_dir: Path, seed: int, stem: ty.Optional[str]):
+def dicom_set_generate_sample_data(dcm_series: DicomSeries, dest_dir: Path, seed: ty.Union[int, Random] = 0, stem: ty.Optional[str] = None) -> ty.Iterable[Path]:
     rng = Random(seed)
     dicom_dir = dicom_dir_generate_sample_data(dcm_series, dest_dir=mkdtemp(), seed=rng, stem=None)[0]
     stem = gen_filename(rng, stem=stem)

extras/fileformats/extras/medimage/diffusion.py

@@ -1,14 +1,14 @@
 import numpy as np
-from fileformats.medimage.diffusion import DwiEncoding, Bval, Bvec
+from fileformats.medimage import DwiEncoding, Bval, Bvec
 
 
 @Bval.read_array.register
-def bval_read_array(bval: Bval):
+def bval_read_array(bval: Bval) -> np.ndarray:  # noqa
     return np.asarray([float(ln) for ln in bval.read_contents().split()])
 
 
 @DwiEncoding.read_array.register
-def bvec_read_array(bvec: Bvec):
+def bvec_read_array(bvec: Bvec) -> np.ndarray:  # noqa
     bvals = bvec.b_values_file.read_array()
     directions = np.asarray(
         [[float(x) for x in ln.split()] for ln in bvec.read_contents().splitlines()]

extras/fileformats/extras/medimage/nifti.py

@@ -1,51 +1,65 @@
 from pathlib import Path
 import typing as ty
+from random import Random
 import nibabel
+import numpy as np
 from fileformats.core import FileSet
+from fileformats.core.utils import gen_filename
 from fileformats.medimage import MedicalImage, Nifti, NiftiGz, Nifti1, NiftiGzX, NiftiX
 import medimages4tests.dummy.nifti
 
 
 @FileSet.read_metadata.register
-def nifti_read_metadata(nifti: Nifti):
+def nifti_read_metadata(nifti: Nifti) -> ty.Mapping[str, ty.Any]:
     return dict(nibabel.load(nifti.fspath).header)
 
 
 @MedicalImage.read_array.register
-def nifti_data_array(nifti: Nifti):
+def nifti_data_array(nifti: Nifti) -> np.ndarray:  # noqa
     return nibabel.load(nifti.fspath).get_data()
 
 
 @MedicalImage.vox_sizes.register
-def nifti_vox_sizes(nifti: Nifti):
+def nifti_vox_sizes(nifti: Nifti) -> ty.Tuple[float, float, float]:
     # FIXME: This won't work for 4-D files
-    return nifti.metadata["pixdim"][1:4]
+    return tuple(float(d) for d in nifti.metadata["pixdim"][1:4])
 
 
 @MedicalImage.dims.register
-def nifti_dims(nifti: Nifti):
+def nifti_dims(nifti: Nifti) -> ty.Tuple[int, int, int]:
     # FIXME: This won't work for 4-D files
-    return nifti.metadata["dim"][1:4]
+    return tuple(int(d) for d in nifti.metadata["dim"][1:4])
 
 
 @FileSet.generate_sample_data.register
-def nifti_generate_sample_data(nifti: Nifti1, dest_dir: Path, seed: int, stem: ty.Optional[str]):
-    return medimages4tests.dummy.nifti.get_image(out_file=dest_dir / "nifti.nii")
+def nifti_generate_sample_data(
+    nifti: Nifti1, dest_dir: Path, seed: ty.Union[int, Random] = 0, stem: ty.Optional[str] = None
+) -> ty.Iterable[Path]:
+    return medimages4tests.dummy.nifti.get_image(
+        out_file=dest_dir / gen_filename(seed, file_type=Nifti1, stem=stem)
+    )
 
 
 @FileSet.generate_sample_data.register
-def nifti_gz_generate_sample_data(nifti: NiftiGz, dest_dir: Path, seed: int, stem: ty.Optional[str]):
+def nifti_gz_generate_sample_data(
+    nifti: NiftiGz, dest_dir: Path, seed: ty.Union[int, Random] = 0, stem: ty.Optional[str] = None
+) -> ty.Iterable[Path]:
     return medimages4tests.dummy.nifti.get_image(
-        out_file=dest_dir / "nifti.nii.gz", compressed=True
+        out_file=dest_dir / gen_filename(seed, file_type=NiftiGz, stem=stem),
+        compressed=True,
     )
 
 
 @FileSet.generate_sample_data.register
-def nifti_gz_x_generate_sample_data(nifti: NiftiGzX, dest_dir: Path, seed: int, stem: ty.Optional[str]):
+def nifti_gz_x_generate_sample_data(
+    nifti: NiftiGzX, dest_dir: Path, seed: ty.Union[int, Random] = 0, stem: ty.Optional[str] = None
+) -> ty.Iterable[Path]:
     return medimages4tests.mri.neuro.t1w.get_image()
 
 
 @FileSet.generate_sample_data.register
-def nifti_x_generate_sample_data(nifti: NiftiX, dest_dir: Path, seed: int, stem: ty.Optional[str]):
+def nifti_x_generate_sample_data(
+    nifti: NiftiX, dest_dir: Path, seed: ty.Union[int, Random] = 0, stem: ty.Optional[str] = None
+) -> ty.Iterable[Path]:
     nifti_gz_x = NiftiGzX(medimages4tests.mri.neuro.t1w.get_image())
     return NiftiX.convert(nifti_gz_x)

fileformats/medimage/__init__.py

@@ -1,20 +1,21 @@
-from ._version import __version__
-from .base import MedicalImage
-from fileformats.application import Dicom  # imported to alias it here as well
+from ._version import __version__  # noqa: F401
+from .base import MedicalImage  # noqa: F401
+# import Dicom to alias to the medimage namespace it here as well
+from fileformats.application import Dicom  # noqa: F401
 from .misc import (  # noqa: F401
     Analyze,
     Mgh,
     MghGz,
 )
-from .nifti import (
+from .nifti import (  # noqa: F401
     Nifti,
     Nifti1,
     Nifti2,
     NiftiGz,
     NiftiX,
     NiftiGzX,
 )
-from .diffusion import (
+from .diffusion import (  # noqa: F401
     DwiEncoding,
     Bvec,
     Bval,
@@ -42,3 +43,6 @@
     Vnd_Siemens_Biograph128Vision_Vr20b_PetCountRate,
     Vnd_Siemens_Biograph128Vision_Vr20b_PetNormalisation,
 )
+from .surface import (
+    Gifti  # noqa: F401
+)

fileformats/medimage/base.py

@@ -19,18 +19,18 @@ class MedicalImage(FileSet):
     binary = True
 
     @hook.extra
-    def read_array(self):
+    def read_array(self) -> "numpy.ndarray":  # noqa
         """
         Returns the binary data of the image in a numpy array
         """
         raise NotImplementedError
 
     @hook.extra
-    def vox_sizes(self) -> ty.Tuple[float]:
+    def vox_sizes(self) -> ty.Tuple[float, float, float]:
         """The length of the voxels along each dimension"""
         raise NotImplementedError
 
     @hook.extra
-    def dims(self) -> ty.Tuple[int]:
+    def dims(self) -> ty.Tuple[int, int, int]:
         """The dimensions of the image"""
         raise NotImplementedError

fileformats/medimage/dicom.py

@@ -26,7 +26,7 @@ def __len__(self):
         return len(self.contents)
 
     @hook.extra
-    def series_number(self):
+    def series_number(self) -> str:
         raise NotImplementedError
 
     @cached_property
@@ -51,7 +51,7 @@ def from_paths(
 
 
 @FileSet.read_metadata.register
-def dicom_collection_read_metadata(collection: DicomCollection) -> ty.Dict[str, ty.Any]:
+def dicom_collection_read_metadata(collection: DicomCollection) -> ty.Mapping[str, ty.Any]:
     # Collated DICOM headers across series
     collated = copy(collection.contents[0].metadata)
     for i, dicom in enumerate(collection.contents[1:], start=1):

fileformats/medimage/diffusion.py

@@ -10,21 +10,21 @@ class DwiEncoding(File):
     iana_mime: ty.Optional[str] = None
 
     @hook.extra
-    def read_array(self):
+    def read_array(self) -> "numpy.ndarray":  # noqa
         "Both the gradient direction and weighting combined into a single Nx4 array"
         raise NotImplementedError
 
     @property
-    def array(self):
+    def array(self) -> "numpy.ndarray":  # noqa
         return self.read_array()
 
     @property
-    def directions(self):
+    def directions(self) -> "numpy.ndarray":  # noqa
         "gradient direction and weighting combined into a single Nx4 array"
         return self.array[:, :3]
 
     @property
-    def b_values(self):
+    def b_values(self) -> "numpy.ndarray":  # noqa
         "the b-value weighting"
         return self.array[:, 3]
 
@@ -34,7 +34,7 @@ class Bval(File):
     ext = ".bval"
 
     @hook.extra
-    def read_array(self):
+    def read_array(self) -> "numpy.ndarray":  # noqa
         raise NotImplementedError
 
 

fileformats/medimage/radlex/__init__.py

@@ -0,0 +1,6 @@
+from fileformats.core import ClassifierCategory
+
+
+class AnatomicalEntity(ClassifierCategory):
+    pass
+

fileformats/medimage/radlex/anatomical/__init__.py

@@ -0,0 +1,5 @@
+from fileformats.core import ClassifierCategory
+
+
+class AnatomicalEntity(ClassifierCategory):
+    pass

fileformats/medimage/radlex/imaging/__init__.py

@@ -0,0 +1,5 @@
+from fileformats.core import ClassifierCategory
+
+
+class ImagingSpecialty(ClassifierCategory):
+    pass