debugging deidentification

extras/fileformats/extras/medimage/__init__.py

@@ -5,3 +5,4 @@
 from . import diffusion
 from . import nifti
 from . import raw
+from . import base

extras/fileformats/extras/medimage/base.py

@@ -6,7 +6,7 @@
 
 
 @extra_implementation(MedicalImage.deidentify)
-def dicom_deidentify(
+def no_deidentification_necessary(
     image: MedicalImage,
     out_dir: ty.Optional[Path] = None,
     new_stem: ty.Optional[str] = None,
@@ -15,8 +15,13 @@ def dicom_deidentify(
     """Assume that no deidentification is needed for medical images by default. We make a
     copy of the image in the output directory for consistency with the behavior of other
     deidentification formats"""
+    if image.contains_phi:
+        raise NotImplementedError(
+            f"{type(image)} images contain Protected Health Information (PHI) and needs a "
+            "specific deidentification method"
+        )
     if out_dir is None:
         out_dir = Path(tempfile.mkdtemp())
     out_dir.mkdir(exist_ok=True, parents=True)
-    image.copy(out_dir, new_stem=new_stem, mode=copy_mode)
-    return image
+    cpy = image.copy(out_dir, new_stem=new_stem, mode=copy_mode)
+    return cpy

extras/fileformats/extras/medimage/dicom.py

@@ -96,7 +96,8 @@ def dicom_deidentify(
         out_dir = Path(tempfile.mkdtemp())
     out_dir.mkdir(parents=True, exist_ok=True)
     dcm = dicom.load()
-    dcm.PatientBirthDate = ""  # dcm.PatientBirthDate[:4] + "0101"
+    dcm.PatientBirthDate = dcm.PatientBirthDate[:4] + "0101"
+    dcm.PatientName = "Anonymous^Anonymous"
     for field in FIELDS_TO_DEIDENTIFY:
         try:
             elem = dcm[field]
@@ -155,7 +156,7 @@ def dicom_collection_deidentify(
     ("0008", "009C"),  # Consulting Physician's Name
     ("0008", "1070"),  # Operators' Name
     ("0010", "4000"),  # Patient Comments
-    ("0010", "0010"),  # Patient's Name
+    # ("0010", "0010"),  # Patient's Name
     ("0010", "0021"),  # Issuer of Patient ID
     ("0010", "0032"),  # Patient's Birth Time
     ("0010", "0050"),  # Patient's Insurance Plan Code SQ

fileformats/medimage/base.py

@@ -2,7 +2,6 @@
 import typing as ty
 from pathlib import Path
 import logging
-from typing_extensions import Self
 from fileformats.core import extra, FileSet, mtime_cached_property
 from fileformats.core.mixin import WithClassifiers
 from .contents import ContentsClassifier
@@ -16,6 +15,10 @@
     from typing import TypeAlias
 else:
     from typing_extensions import TypeAlias
+if sys.version_info >= (3, 12):
+    from typing import Self
+else:
+    from typing_extensions import Self
 
 if ty.TYPE_CHECKING:
     import numpy.typing  # noqa: F401
@@ -39,6 +42,7 @@ class MedicalImage(WithClassifiers, FileSet):
     image_contents = ()
     allowed_classifiers = (ContentsClassifier,)
     exclusive_classifiers = (ImagingModality, AnatomicalEntity, Derivative)
+    contains_phi: bool = True
 
     @extra
     def read_array(self) -> DataArrayType:

fileformats/medimage/dicom.py

@@ -1,14 +1,23 @@
+import sys
 import typing as ty
 from collections import defaultdict, Counter
 from pathlib import Path
-from typing_extensions import Self, TypeAlias
 from fileformats.core.decorators import mtime_cached_property
 from fileformats.core import extra, FileSet, extra_implementation
 from fileformats.core.collection import TypedCollection
 from fileformats.generic import TypedDirectory, TypedSet
 from fileformats.application import Dicom
 from .base import MedicalImage
 
+if sys.version_info >= (3, 9):
+    from typing import TypeAlias
+else:
+    from typing_extensions import TypeAlias
+if sys.version_info >= (3, 12):
+    from typing import Self
+else:
+    from typing_extensions import Self
+
 if ty.TYPE_CHECKING:
     import pydicom.tag
 

fileformats/medimage/misc.py

@@ -14,6 +14,7 @@
 class AnalyzeHeader(BinaryFile):
 
     ext = ".hdr"
+    contains_phi = False
 
 
 class Analyze(WithSeparateHeader, MedicalImage, BinaryFile):

fileformats/medimage/nifti.py

@@ -9,6 +9,7 @@
 class Nifti(MedicalImage, BinaryFile):
 
     ext: str = ".nii"
+    contains_phi = False
 
 
 class WithBids(WithSideCars):

fileformats/medimage/raw/pet/base.py

@@ -1,9 +1,29 @@
+import sys
+import typing as ty
+from pathlib import Path
+from fileformats.core import FileSet, extra
 from fileformats.generic import BinaryFile
 
+if sys.version_info >= (3, 12):
+    from typing import Self
+else:
+    from typing_extensions import Self
+
 
 class PetRawData(BinaryFile):
     """Base class for raw PET data files"""
 
+    @extra
+    def deidentify(
+        self,
+        out_dir: ty.Optional[Path] = None,
+        new_stem: ty.Optional[str] = None,
+        copy_mode: FileSet.CopyMode = FileSet.CopyMode.copy,
+    ) -> Self:
+        """Returns a new copy of the data with any subject-identifying information
+        stripped from the from the data header"""
+        raise NotImplementedError
+
 
 class PetListMode(PetRawData):
     "raw projection data"

pyproject.toml

@@ -7,7 +7,7 @@ name = "fileformats-medimage"
 description = "Classes for representing various medical imaging file formats in Python classes for use in type hinting in data workflows"
 readme = "README.rst"
 requires-python = ">=3.8"
-dependencies = ["fileformats >=0.7.1"]
+dependencies = ["fileformats >=0.14.1"]
 license = { file = "LICENSE" }
 authors = [{ name = "Thomas G. Close", email = "[email protected]" }]
 maintainers = [{ name = "Thomas G. Close", email = "[email protected]" }]
@@ -29,7 +29,14 @@ classifiers = [
 dynamic = ["version"]
 
 [project.optional-dependencies]
-dev = ["black", "pre-commit", "codespell", "flake8", "flake8-pyproject"]
+dev = [
+    "black",
+    "pre-commit",
+    "codespell",
+    "flake8",
+    "flake8-pyproject",
+    "typing_extensions >= 4.6",
+]
 test = [
     "pytest >=6.2.5",
     "pytest-env>=0.6.2",