Fix memory leak by removing custom __copy__ logic

src/torchio/data/subject.py

@@ -4,7 +4,6 @@
 import pprint
 from typing import TYPE_CHECKING
 from typing import Any
-from typing import Callable
 from typing import Dict
 from typing import List
 from typing import Optional
@@ -72,9 +71,6 @@ def __repr__(self):
         )
         return string
 
-    def __copy__(self):
-        return _subject_copy_helper(self, type(self))
-
     def __len__(self):
         return len(self.get_images(intensity_only=False))
 
@@ -429,25 +425,3 @@ def plot(self, **kwargs) -> None:
         from ..visualization import plot_subject  # avoid circular import
 
         plot_subject(self, **kwargs)
-
-
-def _subject_copy_helper(
-    old_obj: Subject,
-    new_subj_cls: Callable[[Dict[str, Any]], Subject],
-):
-    result_dict = {}
-    for key, value in old_obj.items():
-        if isinstance(value, Image):
-            value = copy.copy(value)
-        else:
-            value = copy.deepcopy(value)
-        result_dict[key] = value
-
-    new = new_subj_cls(**result_dict)  # type: ignore[call-arg]
-    new.applied_transforms = old_obj.applied_transforms[:]
-    return new
-
-
-class _RawSubjectCopySubject(Subject):
-    def __copy__(self):
-        return _subject_copy_helper(self, Subject)

src/torchio/datasets/fpg.py

@@ -4,12 +4,12 @@
 from ..data import LabelMap
 from ..data import ScalarImage
 from ..data.io import read_matrix
-from ..data.subject import _RawSubjectCopySubject
+from ..data.subject import Subject
 from ..download import download_url
 from ..utils import get_torchio_cache_dir
 
 
-class FPG(_RawSubjectCopySubject):
+class FPG(Subject):
     """3T :math:`T_1`-weighted brain MRI and corresponding parcellation.
 
     Args:

src/torchio/datasets/itk_snap/itk_snap.py

@@ -2,12 +2,12 @@
 
 from ...data import LabelMap
 from ...data import ScalarImage
-from ...data.subject import _RawSubjectCopySubject
+from ...data.subject import Subject
 from ...download import download_and_extract_archive
 from ...utils import get_torchio_cache_dir
 
 
-class SubjectITKSNAP(_RawSubjectCopySubject):
+class SubjectITKSNAP(Subject):
     """ITK-SNAP Image Data Downloads.
 
     See `the ITK-SNAP website`_ for more information.

src/torchio/datasets/mni/mni.py

@@ -1,8 +1,8 @@
-from ...data.subject import _RawSubjectCopySubject
+from ...data.subject import Subject
 from ...utils import get_torchio_cache_dir
 
 
-class SubjectMNI(_RawSubjectCopySubject):
+class SubjectMNI(Subject):
     """Atlases from the Montreal Neurological Institute (MNI).
 
     See `the website <https://nist.mni.mcgill.ca/?page_id=714>`_ for more

src/torchio/datasets/slicer.py

@@ -1,7 +1,7 @@
 import urllib.parse
 
 from ..data import ScalarImage
-from ..data.subject import _RawSubjectCopySubject
+from ..data.subject import Subject
 from ..download import download_url
 from ..utils import get_torchio_cache_dir
 
@@ -36,7 +36,7 @@
 }
 
 
-class Slicer(_RawSubjectCopySubject):
+class Slicer(Subject):
     """Sample data provided by `3D Slicer <https://www.slicer.org/>`_.
 
     See `the Slicer wiki <https://www.slicer.org/wiki/SampleData>`_

src/torchio/transforms/transform.py

@@ -158,9 +158,9 @@ def __call__(self, data: InputType) -> InputType:
         if self.keep is not None:
             images_to_keep = {}
             for name, new_name in self.keep.items():
-                images_to_keep[new_name] = copy.copy(subject[name])
+                images_to_keep[new_name] = copy.deepcopy(subject[name])
         if self.copy:
-            subject = copy.copy(subject)
+            subject = copy.deepcopy(subject)
         with np.errstate(all='raise', under='ignore'):
             transformed = self.apply_transform(subject)
         if self.keep is not None:

tests/data/test_subject.py

@@ -186,3 +186,30 @@ def test_subjects_batch(self):
         loader = tio.SubjectsLoader(subjects, batch_size=4)
         batch = next(iter(loader))
         assert batch.__class__ is dict
+
+    def test_deep_copy_subject(self):
+        sub_copy = copy.deepcopy(self.sample_subject)
+        assert isinstance(sub_copy, tio.data.Subject)
+
+        new_tensor = torch.ones_like(sub_copy['t1'].data)
+        sub_copy['t1'].set_data(new_tensor)
+        # The data of the original subject should not be modified
+        assert not torch.allclose(sub_copy['t1'].data, self.sample_subject['t1'].data)
+
+    def test_shallow_copy_subject(self):
+        # We are creating a deep copy of the original subject first to not modify the original subject
+        copy_original_subj = copy.deepcopy(self.sample_subject)
+        sub_copy = copy.copy(copy_original_subj)
+        assert isinstance(sub_copy, tio.data.Subject)
+
+        new_tensor = torch.ones_like(sub_copy['t1'].data)
+        sub_copy['t1'].set_data(new_tensor)
+
+        # The data of both copies needs to be the same as we are using a shallow copy
+        assert torch.allclose(sub_copy['t1'].data, copy_original_subj['t1'].data)
+        # The data of the original subject should not be modified
+        assert not torch.allclose(
+            sub_copy['t1'].data, self.sample_subject['t1'].data
+        ) and not torch.allclose(
+            copy_original_subj['t1'].data, self.sample_subject['t1'].data
+        )