feat(repersonlization): add customizable fields (#59)

* feat: add customize repersonalization fields

* update readme

* Added file with accession number

* Add test for custom fields

---------

Co-authored-by: Ashley Rudelsheim <[email protected]>
Co-authored-by: Ashley Rudelsheim <[email protected]>
Co-authored-by: Ashley Rudelsheim <[email protected]>

Author: a-parida12

README.md

@@ -93,7 +93,7 @@ Returns:
 
 ## Repersonalisation
 
-It is the process of copying over data regarding the identity of the encapsualted pdf from a template dicom. Currently, the fileds that are repersonalised are-
+It is the process of copying over data regarding the identity of the encapsualted pdf from a template dicom. Currently, the fields that are repersonalised by default are-
 
 - PatientName
 - PatientID
@@ -103,3 +103,20 @@ It is the process of copying over data regarding the identity of the encapsualte
 - ~~SOPInstanceUID~~
 
 The fields `SeriesInstanceUID` and `SOPInstanceUID` have been removed from the repersonalization by copying as it violates the DICOM standards.
+
+You can set the fields to repersonalize by passing repersonalisation_fields into `Pdf2EncapsDCM()`, or `Pdf2RgbSC()`
+
+Example:
+
+```python
+fields = [
+    "PatientName",
+    "PatientID",
+    "PatientSex",
+    "StudyInstanceUID",
+    "AccessionNumber"
+]
+converter = Pdf2RgbSC(repersonalisation_fields=fields)
+```
+
+note: this will overwrite the default fields.

pdf2dcm/base.py

@@ -14,13 +14,16 @@
 
 
 class BaseConverter(ABC):
-    def __init__(self):
-        self.repersonalisation_fields = [
-            "PatientName",
-            "PatientID",
-            "PatientSex",
-            "StudyInstanceUID",
-        ]
+    def __init__(self, repersonalisation_fields=[]):
+        if len(repersonalisation_fields):
+            self.repersonalisation_fields = repersonalisation_fields
+        else:
+            self.repersonalisation_fields = [
+                "PatientName",
+                "PatientID",
+                "PatientSex",
+                "StudyInstanceUID",
+            ]
 
     def personalize_dcm(
         self, template_dcm_path: Path, pdf_dcm: FileDataset

pdf2dcm/pdf2encaps.py

@@ -7,9 +7,13 @@
 
 
 class Pdf2EncapsDCM(BaseConverter):
-    def __init__(self):
-        """Class for Encapsulated PDF generation"""
-        super().__init__()
+    def __init__(self, repersonalisation_fields=[]):
+        """Class for Encapsulated PDF generation
+        
+        Args:
+            repersonalisation_fields (List<String>, optional): fields to be copied to the new image
+        """
+        super().__init__(repersonalisation_fields=repersonalisation_fields)
 
     def _get_encapspdf_meta(self) -> FileMetaDataset:
         """Get and set the file meta information for the pdf encaps dicom

pdf2dcm/pdf2rgb.py

@@ -11,16 +11,18 @@
 
 
 class Pdf2RgbSC(BaseConverter):
-    def __init__(self, dpi=144, merge_pages=False):
+    def __init__(self, dpi=144, merge_pages=False, repersonalisation_fields=[]):
         """Class for the generation RGB Secondary capture
 
         Args:
             dpi (int, optional): dots per inch, set resolution of the image. Defaults to 144.
             merge_pages (bool, optional): multiple pgs must be put into 1 dicom. Defaults to False.
+            repersonalisation_fields (List<String>, optional): fields to be copied to the new image
+
         """
         self.merge_pages_flag = merge_pages
         self.dpi = dpi
-        super().__init__()
+        super().__init__(repersonalisation_fields=repersonalisation_fields)
 
     def _get_rgbsc_meta(self) -> FileMetaDataset:
         """Get and set the file meta information for the rgb secondary capture dicom

tests/conftest.py

@@ -23,3 +23,13 @@ def pdfencapsconverter():
 @pytest.fixture
 def rgbscconverter():
     yield Pdf2RgbSC()
+
+@pytest.fixture
+def pdfrepersonconverter():
+    yield Pdf2EncapsDCM(repersonalisation_fields=[
+        "PatientName",
+        "PatientID",
+        "PatientSex",
+        "StudyInstanceUID",
+        "AccessionNumber"
+    ])

tests/test_03_repersonalisation.py

@@ -71,3 +71,26 @@ def test_03_2_name_missing(pdfencapsconverter):
     assert dcm_ds.PatientName == ""
 
     os.remove(stored_path)
+
+@pytest.mark.reperson
+def test_03_4_additional_fields_personlisation(pdfrepersonconverter):
+    path_pdf = "tests/test_data/test_file.pdf"
+    ref_dicom = "tests/test_data/CT_small_accession_number.dcm"
+
+    # with personalisation
+    stored_path = pdfrepersonconverter.run(path_pdf, ref_dicom)[0]
+
+    assert os.path.exists(stored_path)
+    assert pdfrepersonconverter.check_valid_dcm(stored_path)
+
+    dcm_ds = pydicom.dcmread(stored_path)
+    ref_dcm_ds = pydicom.dcmread(ref_dicom)
+
+    # check repersonaliation attribute
+    assert len(dcm_ds.EncapsulatedDocument) == 898332
+    assert dcm_ds.PatientName == ref_dcm_ds.PatientName
+    assert dcm_ds.PatientID == ref_dcm_ds.PatientID
+    assert dcm_ds.PatientSex == ref_dcm_ds.PatientSex
+    assert dcm_ds.AccessionNumber == ref_dcm_ds.AccessionNumber
+
+    os.remove(stored_path)