ENH+CMP: Speed up selective extraction, move to pydicom.dcmread

moloney · moloney · commit c0733d24876c · 2024-12-03T17:34:40.000-08:00
diff --git a/src/dcmstack/dcmstack.py b/src/dcmstack/dcmstack.py
@@ -11,10 +11,8 @@
     from ordereddict import OrderedDict
 
 import numpy as np
-try:
-    import pydicom
-except ImportError:
-    import dicom as pydicom
+import pydicom
+from pydicom.datadict import tag_for_keyword
 import nibabel as nb
 from nibabel.nifti1 import Nifti1Extensions
 from nibabel.spatialimages import HeaderDataError
@@ -353,12 +351,13 @@ def get_ordinate(self, nw):
 '''
 
 
-_pix_attrs = ('PixelData', 'FloatPixelData', 'DoubleFloatPixelData')
+_pix_tags = tuple(tag_for_keyword(x) for x in ('PixelData', 'FloatPixelData', 'DoubleFloatPixelData'))
 
 
 def is_image(dcm):
     '''Test if the data set is an image'''
-    if any(hasattr(dcm, attr) for attr in _pix_attrs):
+    tags = set(dcm.keys())
+    if any(t in tags for t in _pix_tags):
         return True
     return False
 
@@ -1056,7 +1055,7 @@ def parse_and_group(src_paths, group_by=DEFAULT_GROUP_KEYS, extractor=None,
     for dcm_path in src_paths:
         #Read the DICOM file
         try:
-            dcm = pydicom.read_file(dcm_path, force=force)
+            dcm = pydicom.dcmread(dcm_path, force=force, defer_size=32)
         except Exception as e:
             if warn_on_except:
                 warnings.warn('Error reading file %s: %s' % (dcm_path, str(e)))
diff --git a/src/dcmstack/extract.py b/src/dcmstack/extract.py
@@ -11,7 +11,7 @@
 import pydicom
 from pydicom.tag import BaseTag
 from pydicom.dataset import PrivateBlock
-from pydicom.datadict import keyword_for_tag, private_dictionaries
+from pydicom.datadict import keyword_for_tag, tag_for_keyword, private_dictionaries
 from pydicom.charset import decode_element
 from nibabel.nicom import csareader
 
@@ -76,10 +76,9 @@ def ignore_unknown_private(tag, name, ds):
 
 def make_ignore_except_rule(include):
     """Make rule that ignores everying not in `include`"""
+    incl_tags = set([tag_for_keyword(x) for x in include if isinstance(x, str)])
     def ignore_except(tag, name, ds):
-        if tag in include:
-            return False
-        return name not in include
+        return tag not in incl_tags
     return ignore_except
 
 
diff --git a/test/test_dcmmeta.py b/test/test_dcmmeta.py
@@ -1039,7 +1039,7 @@ def test_from_dicom():
                          'dcmstack',
                          '2D_16Echo_qT2')
     src_fn = path.join(data_dir, 'TE_40_SlcPos_-33.707626341697.dcm')
-    src_dcm = pydicom.read_file(src_fn)
+    src_dcm = pydicom.dcmread(src_fn)
     src_dw = nb.nicom.dicomwrappers.wrapper_from_data(src_dcm)
     meta = {'EchoTime': 40}
     nw = dcmmeta.NiftiWrapper.from_dicom(src_dcm, meta)
diff --git a/test/test_dcmstack.py b/test/test_dcmstack.py
@@ -205,7 +205,7 @@ def test_image_collision():
                          'dcmstack',
                          '2D_16Echo_qT2',
                          'TE_20_SlcPos_-33.707626341697.dcm')
-    dcm = pydicom.read_file(dcm_path)
+    dcm = pydicom.dcmread(dcm_path)
     stack = dcmstack.DicomStack('EchoTime')
     stack.add_dcm(dcm)
     with pytest.raises(dcmstack.ImageCollisionError):
@@ -259,7 +259,7 @@ def setup_method(self, method):
                              'data',
                              'dcmstack',
                              '2D_16Echo_qT2')
-        self.inputs = [pydicom.read_file(path.join(data_dir, fn))
+        self.inputs = [pydicom.dcmread(path.join(data_dir, fn))
                        for fn in ('TE_20_SlcPos_-33.707626341697.dcm',
                                   'TE_20_SlcPos_-23.207628249046.dcm',
                                   'TE_40_SlcPos_-33.707626341697.dcm',
@@ -310,7 +310,7 @@ def setup_method(self, method):
                              'data',
                              'dcmstack',
                              '2D_16Echo_qT2')
-        self.inputs = [pydicom.read_file(path.join(data_dir, fn))
+        self.inputs = [pydicom.dcmread(path.join(data_dir, fn))
                        for fn in ('TE_40_SlcPos_-33.707626341697.dcm',
                                   'TE_40_SlcPos_-23.207628249046.dcm',
                                   'TE_60_SlcPos_-33.707626341697.dcm',
@@ -355,7 +355,7 @@ def setup_method(self, method):
                              'data',
                              'dcmstack',
                              '2D_16Echo_qT2')
-        self.inputs = [pydicom.read_file(path.join(data_dir, fn))
+        self.inputs = [pydicom.dcmread(path.join(data_dir, fn))
                        for fn in ('TE_40_SlcPos_-33.707626341697.dcm',
                                   'TE_40_SlcPos_-23.207628249046.dcm',
                                   'TE_60_SlcPos_-33.707626341697.dcm',
@@ -408,7 +408,7 @@ def setup_method(self, method):
                              'data',
                              'dcmstack',
                              '2D_16Echo_qT2')
-        self.inputs = [pydicom.read_file(path.join(data_dir, fn))
+        self.inputs = [pydicom.dcmread(path.join(data_dir, fn))
                        for fn in ('TE_40_SlcPos_-33.707626341697.dcm',
                                   'TE_40_SlcPos_-23.207628249046.dcm',
                                   'TE_60_SlcPos_-33.707626341697.dcm',
@@ -477,7 +477,7 @@ def setup_method(self, method):
                              'data',
                              'dcmstack',
                              '2D_16Echo_qT2')
-        self.inputs = [pydicom.read_file(path.join(self.data_dir, fn))
+        self.inputs = [pydicom.dcmread(path.join(self.data_dir, fn))
                        for fn in ('TE_20_SlcPos_-33.707626341697.dcm',
                                   'TE_20_SlcPos_-23.207628249046.dcm'
                                  )
@@ -544,7 +544,7 @@ def setup_method(self, method):
                              'data',
                              'dcmstack',
                              '2D_16Echo_qT2')
-        self.inputs = [pydicom.read_file(path.join(self.data_dir, fn))
+        self.inputs = [pydicom.dcmread(path.join(self.data_dir, fn))
                        for fn in ('TE_20_SlcPos_-33.707626341697.dcm',
                                   'TE_20_SlcPos_-23.207628249046.dcm',
                                   'TE_40_SlcPos_-33.707626341697.dcm',
@@ -646,7 +646,7 @@ def setup_method(self, method):
     def test_default(self):
         res = dcmstack.parse_and_group(self.in_paths)
         assert len(res) == 1
-        ds = pydicom.read_file(self.in_paths[0])
+        ds = pydicom.dcmread(self.in_paths[0])
         group_key = list(res.keys())[0]
         for attr_idx, attr in enumerate(dcmstack.DEFAULT_GROUP_KEYS):
             if attr in dcmstack.DEFAULT_CLOSE_KEYS:
@@ -672,7 +672,7 @@ def setup_method(self, method):
     def test_default(self):
         res = dcmstack.parse_and_stack(self.in_paths)
         assert len(res) == 1
-        ds = pydicom.read_file(self.in_paths[0])
+        ds = pydicom.dcmread(self.in_paths[0])
         group_key = list(res.keys())[0]
         for attr_idx, attr in enumerate(dcmstack.DEFAULT_GROUP_KEYS):
             if attr in dcmstack.DEFAULT_CLOSE_KEYS:
diff --git a/test/test_extract.py b/test/test_extract.py
@@ -20,7 +20,7 @@
 class TestCsa(object):
     def setup_method(self, method):
         data_fn = path.join(test_dir, 'data', 'extract', 'csa_test.dcm')
-        self.data = pydicom.read_file(data_fn)
+        self.data = pydicom.dcmread(data_fn)
 
     def teardown_method(self, method):
         del self.data
@@ -70,7 +70,7 @@ def test_csa_series_trans(self):
 class TestMetaExtractor(object):
     def setup_method(self, method):
         data_fn = path.join(test_dir, 'data', 'extract', 'csa_test.dcm')
-        self.data = pydicom.read_file(data_fn)
+        self.data = pydicom.dcmread(data_fn)
 
     def teardown_method(self,method):
         del self.data
