Add tests for keep_all_tags option.

Fix pydicom value represetations to string.

rap_sitkcore/_dicom_util.py

@@ -24,7 +24,7 @@ def convert_mv_ds_to_float_list(rep: str, vm: int = 0) -> List[float]:
 
 def convert_float_list_to_mv_ds(value: List[float]) -> str:
     """
-    Convert a iterable of float to the DICOM mutli-value representation for decimal string (DS).
+    Convert a iterable of float to the DICOM multi-value representation for decimal string (DS).
 
     This method is intended to convert the pydicom MV DS data elements to the representation that GDCM produced for
     SimpleITK.
@@ -42,6 +42,26 @@ def convert_float_list_to_mv_ds(value: List[float]) -> str:
     return rep
 
 
+def convert_int_list_to_mv_ds(value: List[float]) -> str:
+    """
+    Convert a iterable of int to the DICOM multi-value representation for (unsigned) integer (US/IS).
+
+    This method is intended to convert the pydicom MV DS data elements to the representation that GDCM produced for
+    SimpleITK.
+
+    :param value: an iterable or list like object of convertable to float values.
+    :returns: The value encode in for DICOM representation.
+    """
+
+    rep = _vm_delimiter.join([str(int(f)) for f in value])
+
+    # DICOM spec
+    if len(rep) % 2:
+        rep += " "
+
+    return rep
+
+
 def keyword_to_gdcm_tag(keyword: str) -> str:
     """Converts a DICOM keyword to a DICOM tag formatted as a string to match GDCM representation.
 

rap_sitkcore/read_dcm.py

@@ -2,7 +2,7 @@
 import pydicom
 from pathlib import Path
 from rap_sitkcore._util import srgb2gray
-from rap_sitkcore._dicom_util import convert_float_list_to_mv_ds, keyword_to_gdcm_tag
+from rap_sitkcore._dicom_util import convert_float_list_to_mv_ds, convert_int_list_to_mv_ds, keyword_to_gdcm_tag
 import logging
 
 _logger = logging.getLogger(__name__)
@@ -37,10 +37,16 @@ def _get_string_representation(de: pydicom.dataelem.DataElement) -> str:
                 return convert_float_list_to_mv_ds(de.value)
             else:
                 return str(float(de.value))
-        elif de.VR == "US":
-            return str(int(de.value))
+        elif de.VR in ["US", "IS"]:
+
+            if de.VM > 1:
+                return convert_int_list_to_mv_ds(de.value)
+            else:
+                assert str(int(de.value)) == str(de.value), f"{de.value} != {int(de.value)}"
+                return str(int(de.value))
+
         else:
-            return de.value
+            return str(de.value)
     except (TypeError, ValueError) as e:
         raise RuntimeError(
             f'"Error parsing data element "{de.name}" with value "{de.value}" '
@@ -66,6 +72,7 @@ def _read_dcm_pydicom(filename: Path, keep_all_tags: bool = False) -> sitk.Image
     elif ds.PhotometricInterpretation in ["YBR_FULL_422", "YBR_FULL", "RGB"]:
         if ds.PhotometricInterpretation != "RGB":
             from pydicom.pixel_data_handlers.util import convert_color_space
+
             arr = convert_color_space(ds.pixel_array, ds.PhotometricInterpretation, "RGB")
 
         img = sitk.GetImageFromArray(arr, isVector=True)
@@ -77,6 +84,7 @@ def _read_dcm_pydicom(filename: Path, keep_all_tags: bool = False) -> sitk.Image
         for de in ds:
             if de.keyword != "PixelData":
                 key = f"{de.tag.group:04x}|{de.tag.elem:04x}"
+                # print(f"pydicom tag: {key} = \"{de.value}\" type: {type(de.value)} VR: {de.VR} VM: {de.VM}")
                 img[key] = _get_string_representation(de)
     # iterate through all tags and copy the ones specified in _keyword_to_copy
     # to the SimpleITK image
@@ -152,14 +160,15 @@ def read_dcm(filename: Path, keep_all_tags: bool = False) -> sitk.Image:
     if img.GetNumberOfComponentsPerPixel() == 1:
         out = img
     elif img.GetNumberOfComponentsPerPixel() == 3:
-
         out = srgb2gray(img)
 
-        # After converting to grayscale, we need to copy the tags from the original image
+        # Copy all tags
         old_keys = img.GetMetaDataKeys()
-        if keep_all_tags:
-            for k in old_keys:
-                out[k] = img[k]
+
+        for k in old_keys:
+            out[k] = img[k]
+    else:
+        raise RuntimeError(f"Unsupported number of components: {img.GetNumberOfComponentsPerPixel()}")
 
     if not keep_all_tags:
         old_keys = set(out.GetMetaDataKeys())
@@ -168,5 +177,3 @@ def read_dcm(filename: Path, keep_all_tags: bool = False) -> sitk.Image:
             del out[k]
 
     return out
-
-    raise RuntimeError(f"Unsupported number of components: {img.GetNumberOfComponentsPerPixel()}")

test/unit/test_read_dcm.py

@@ -66,6 +66,74 @@ def test_read_dcm1(test_file, data_paths):
         assert k in _white_listed_dicom_tags
 
 
+@pytest.mark.parametrize(
+    "test_file,number_of_tags",
+    [
+        ("1.3.6.1.4.1.25403.163683357445804.11044.20131119114627.12.dcm", 109),
+        ("1.3.6.1.4.1.25403.158515237678667.5060.20130807021253.18.dcm", 33),
+        ("1.2.840.114062.2.192.168.196.13.2015.11.4.13.11.45.13871156.dcm", 37),
+        ("2.25.288816364564751018524666516362407260298.dcm", 15),
+        ("2.25.240995260530147929836761273823046959883.dcm", 15),
+        ("2.25.226263219114459199164755074787420926696.dcm", 15),
+        ("2.25.40537326380965754670062689705190363681.dcm", 15),
+        ("2.25.326714092011492114153708980185182745084.dcm", 15),
+        ("2.25.5871713374023139953558641168991505875.dcm", 15),
+        ("n10.dcm", 30),
+        ("n11.dcm", 30),
+        ("n12.dcm", 30),
+        ("1.2.392.200036.9116.2.5.1.37.2429823676.1495586039.603772.DCM", 94),
+        ("2.25.298570032897489859462791131067889681111.dcm", 15),
+        ("non_square_color.dcm", 15),
+        ("non_square_uint16.dcm", 57),
+        ("square_uint8.dcm", 32),
+    ],
+)
+def test_read_dcm_pydicom1(test_file, number_of_tags, data_paths):
+    filename = data_paths[test_file]
+
+    required_tags = [
+        "StudyInstanceUID",
+        "SeriesInstanceUID",
+        "Modality",
+    ]
+
+    img = _read_dcm_pydicom(Path(filename))
+    for tag in required_tags:
+        key = keyword_to_gdcm_tag(tag)
+        assert key in img
+
+    for k in img.GetMetaDataKeys():
+        assert k in _white_listed_dicom_tags
+
+    img = _read_dcm_pydicom(Path(filename), keep_all_tags=True)
+
+    img_keys = set(img.GetMetaDataKeys())
+
+    for tag in required_tags:
+        key = keyword_to_gdcm_tag(tag)
+        assert key in img
+
+    # Check that
+    assert (
+        len(img_keys - set(_white_listed_dicom_tags)) == number_of_tags
+    ), f"Expected: {number_of_tags} but got {len(img_keys - set(_white_listed_dicom_tags))}"
+
+    img = rap_sitkcore.read_dcm(Path(filename), keep_all_tags=True)
+
+    img_keys = set(img.GetMetaDataKeys())
+
+    for tag in required_tags:
+        key = keyword_to_gdcm_tag(tag)
+        assert key in img
+
+    # There are 1-4 different number tags between pydicom and ITK GDCM
+    # assert len (img_keys - set(_white_listed_dicom_tags)) == number_of_tags+2,\
+    #    f"Expected: {number_of_tags} but got {len(img_keys - set(_whihoyte_listed_dicom_tags))}"
+
+    # BUG: Not all these file can be written out
+    # sitk.WriteImage(img, "foo.dcm")
+
+
 def test_read_dcm2():
     """Test with filename does not exit"""