🚸 Handle Multidimensional Features in WSIReader (#742)

- Adds support for multichannel inputs e.g., image masks, feature maps etc. Fixes #609 and #610 

---------

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Shan E Ahmed Raza <13048456+shaneahmed@users.noreply.github.com>

tests/test_wsireader.py

@@ -81,6 +81,7 @@
 }
 RNG = np.random.default_rng()  # Numpy Random Generator
 
+
 # -------------------------------------------------------------------------------------
 # Utility Test Functions
 # -------------------------------------------------------------------------------------
@@ -2110,7 +2111,6 @@ def test_store_reader_alpha(remote_sample: Callable) -> None:
         wsi_reader.info,
         base_wsi=wsi_reader,
     )
-    store_reader.renderer.info["mpp"] = store_reader.info.as_dict()["mpp"]
     wsi_thumb = wsi_reader.slide_thumbnail()
     wsi_tile = wsi_reader.read_rect((500, 500), (1000, 1000))
     store_thumb = store_reader.slide_thumbnail()
@@ -2759,3 +2759,34 @@ def test_file_path_does_not_exist() -> None:
 def test_read_mpp(wsi: WSIReader) -> None:
     """Test that the mpp is read correctly."""
     assert wsi.info.mpp == pytest.approx(0.25, 1)
+
+
+def test_read_multi_channel(source_image: Path) -> None:
+    """Test reading image with more than three channels.
+
+    Create a virtual WSI by concatenating the source_image.
+
+    """
+    img_array = utils.misc.imread(Path(source_image))
+    new_img_array = np.concatenate((img_array, img_array), axis=-1)
+
+    new_img_size = new_img_array.shape[:2][::-1]
+    meta = wsireader.WSIMeta(slide_dimensions=new_img_size, axes="YXS", mpp=(0.5, 0.5))
+    wsi = wsireader.VirtualWSIReader(new_img_array, info=meta)
+
+    region = wsi.read_rect(
+        location=(0, 0),
+        size=(50, 100),
+        pad_mode="reflect",
+        units="mpp",
+        resolution=0.25,
+    )
+    target = cv2.resize(
+        new_img_array[:50, :25, :],
+        (50, 100),
+        interpolation=cv2.INTER_CUBIC,
+    )
+
+    assert region.shape == (100, 50, (new_img_array.shape[-1]))
+    assert np.abs(np.median(region.astype(int) - target.astype(int))) == 0
+    assert np.abs(np.mean(region.astype(int) - target.astype(int))) < 0.2

tiatoolbox/wsicore/wsireader.py

@@ -2792,8 +2792,14 @@ class VirtualWSIReader(WSIReader):
     :func:`~tiatoolbox.utils.image.sub_pixel_read`.
 
     Attributes:
-        img (:class:`numpy.ndarray`)
-        mode (str)
+        img (:class:`numpy.ndarray`):
+            Input image as :class:`numpy.ndarray`.
+        mode (str):
+            Mode of the input image. Default is 'rgb'. Allowed values
+            are: rgb, bool, feature. "rgb" mode supports bright-field color images.
+            "bool" mode supports binary masks,
+            interpolation in this case will be "nearest" instead of "bicubic".
+            "feature" mode allows multichannel features.
 
     Args:
         input_img (str, :obj:`Path`, :class:`numpy.ndarray`):
@@ -2802,7 +2808,10 @@ class VirtualWSIReader(WSIReader):
             Metadata for the virtual wsi.
         mode (str):
             Mode of the input image. Default is 'rgb'. Allowed values
-            are: rgb, bool.
+            are: rgb, bool, feature. "rgb" mode supports bright-field color images.
+            "bool" mode supports binary masks,
+            interpolation in this case will be "nearest" instead of "bicubic".
+            "feature" mode allows multichannel features.
 
     """
 
@@ -2820,15 +2829,26 @@ def __init__(
             mpp=mpp,
             power=power,
         )
-        if mode.lower() not in ["rgb", "bool"]:
+        if mode.lower() not in ["rgb", "bool", "feature"]:
             msg = "Invalid mode."
             raise ValueError(msg)
-        self.mode = mode.lower()
+
         if isinstance(input_img, np.ndarray):
             self.img = input_img
         else:
             self.img = utils.imread(self.input_path)
 
+        if mode != "bool" and (
+            self.img.ndim == 2 or self.img.shape[2] not in [3, 4]  # noqa: PLR2004
+        ):
+            logger.warning(
+                "The image mode is set to 'feature' as the input"
+                " dimensions do not match with binary mask or RGB/RGBA.",
+            )
+            mode = "feature"
+
+        self.mode = mode.lower()
+
         if info is not None:
             self._m_info = info
 
@@ -3278,6 +3298,9 @@ class docstrings for more information.
         if interpolation in [None, "none"]:
             interpolation = None
 
+        if interpolation == "optimise" and self.mode == "bool":
+            interpolation = "nearest"
+
         im_region = utils.image.sub_pixel_read(
             self.img,
             bounds_at_read,