Merge pull request #952 from NHSDigital/fix/save-valid-jpeg-from-dicom

FIX: Save valid jpeg from DICOM pixel data

Author: steventux

.gitignore

@@ -39,3 +39,6 @@ coverage
 .structurizr
 workspace.json
 pyrightconfig.json
+
+# API local file uploads containing DICOM data
+manage_breast_screening/data/dicom

manage_breast_screening/config/settings.py

@@ -209,6 +209,9 @@ def list_env(key):
         # In non-production environments without a connection string, use local file storage
         dicom_storage_options = {
             "BACKEND": "django.core.files.storage.FileSystemStorage",
+            "OPTIONS": {
+                "location": BASE_DIR / "data" / "dicom",
+            },
         }
 else:
     # In production, use DefaultAzureCredential for authentication

manage_breast_screening/dicom/dicom_recorder.py

@@ -1,7 +1,9 @@
 import io
 from datetime import datetime
 
+import numpy as np
 import pydicom
+from django.core.files.uploadedfile import InMemoryUploadedFile
 from PIL import Image as PILImage
 
 from .models import Image, Series, Study
@@ -46,8 +48,7 @@ def get_or_create_records(
         if created:
             image.dicom_file.save(f"{sop_uid}.dcm", dicom_file)
             image.image_file.save(
-                f"{sop_uid}.jpeg",
-                io.BytesIO(PILImage.fromarray(ds.pixel_array).tobytes()),
+                f"{sop_uid}.jpeg", __class__.dataset_to_jpeg(sop_uid, ds)
             )
 
         return study, series, image
@@ -61,3 +62,27 @@ def study_date_and_time(ds) -> datetime | None:
                 study_date + study_time.split(".")[0], "%Y%m%d%H%M%S"
             )
         return None
+
+    @staticmethod
+    def dataset_to_jpeg(sop_uid: str, ds: pydicom.Dataset) -> InMemoryUploadedFile:
+        """Convert a DICOM dataset to a JPEG image and return it as an InMemoryUploadedFile."""
+        # Normalize pixel data to 0-255 and convert to uint8
+        pixel_array = ds.pixel_array
+        pixel_array = pixel_array.astype(np.float32)
+        pixel_array -= pixel_array.min()
+        pixel_array /= pixel_array.max()
+        pixel_array *= 255.0
+        pixel_array = pixel_array.astype(np.uint8)
+        # Create a PIL image from the pixel array
+        image = PILImage.fromarray(pixel_array, mode="L")
+        in_memory_file = io.BytesIO()
+        image.save(in_memory_file, format="JPEG")
+
+        return InMemoryUploadedFile(
+            in_memory_file,
+            name=f"{sop_uid}.jpeg",
+            field_name="file",
+            content_type="image/jpeg",
+            size=in_memory_file.getbuffer().nbytes,
+            charset=None,
+        )

manage_breast_screening/dicom/tests/test_dicom_recorder.py

@@ -1,6 +1,8 @@
 import tempfile
 from datetime import datetime
+from unittest.mock import patch
 
+import numpy as np
 import pydicom
 import pytest
 
@@ -84,3 +86,18 @@ def test_get_or_create_records_invalid_dicom(self, source_message_id):
             with open(temp_file.name, "rb") as dicom_file:
                 with pytest.raises(AttributeError):
                     DicomRecorder.get_or_create_records(source_message_id, dicom_file)
+
+    def test_dataset_to_jpeg(self, dataset):
+        converted_pixel_array = dataset.pixel_array.astype(np.float32)
+        converted_pixel_array -= converted_pixel_array.min()
+        converted_pixel_array /= converted_pixel_array.max()
+        converted_pixel_array *= 255.0
+        converted_pixel_array = converted_pixel_array.astype(np.uint8)
+
+        with patch(f"{DicomRecorder.__module__}.PILImage.fromarray") as mock_fromarray:
+            DicomRecorder.dataset_to_jpeg(dataset.SOPInstanceUID, dataset)
+
+            assert mock_fromarray.call_count == 1
+            assert mock_fromarray.call_args[0][0].shape == converted_pixel_array.shape
+            assert np.array_equal(mock_fromarray.call_args[0][0], converted_pixel_array)
+            assert mock_fromarray.call_args[1]["mode"] == "L"