Correctly account for Cr and Cb subsampling for native YBR_FULL_422

src/highdicom/io.py

@@ -492,6 +492,12 @@ def _bytes_per_frame_uncompressed(self) -> int:
             #   are packed into 12.5 -> 13 bytes
             return n_pixels // 8 + (n_pixels % 8 > 0)
         else:
+            if self.metadata.PhotometricInterpretation == 'YBR_FULL_422':
+                # Account for subsampling of CB and CR when calculating
+                # expected number of samples
+                # See https://dicom.nema.org/medical/dicom/current/output/chtml
+                # /part03/sect_C.7.6.3.html#sect_C.7.6.3.1.2
+                n_pixels = self.metadata.Rows * self.metadata.Columns * 2
             return n_pixels * bits_allocated // 8
 
     def close(self) -> None:

tests/test_io.py

@@ -157,6 +157,25 @@ def test_read_multi_frame_seg_image_sm_dots_bitpacked(self):
                 )
                 np.testing.assert_array_equal(frame, pixel_array[i, ...])
 
+    def test_read_ybr_422_native(self):
+        # Reading a frame using YBR_422 photometric interpretation and no
+        # compression
+        filename = str(get_testdata_file('SC_ybr_full_422_uncompressed.dcm'))
+        dataset = dcmread(filename)
+        pixel_array = dataset.pixel_array
+        with ImageFileReader(filename) as reader:
+            assert reader.number_of_frames == 1
+            frame = reader.read_frame(0, correct_color=False)
+            assert isinstance(frame, np.ndarray)
+            assert frame.ndim == 3
+            assert frame.dtype == np.uint8
+            assert frame.shape == (
+                reader.metadata.Rows,
+                reader.metadata.Columns,
+                reader.metadata.SamplesPerPixel,
+            )
+            np.testing.assert_array_equal(frame, pixel_array)
+
     def test_read_single_frame_ct_image_dicom_bytes_io(self):
         filename = str(self._test_dir.joinpath("ct_image.dcm"))
         dcm = DicomBytesIO(open(filename, "rb").read())