🐛 Fix TIFFWSIReader read_bound (#777)

![image](https://github.com/TissueImageAnalytics/tiatoolbox/assets/24943262/20c0fda0-19f6-46cb-97f8-9ab125d3b3be)

Emergency bugfix per @John-P request.
The culprit is reading bound doesn't use the adjusted bounds as have been done in OpenSlideReader.

---------

Co-authored-by: Abdol <a@fkrtech.com>
Co-authored-by: measty <20169086+measty@users.noreply.github.com>
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/conftest.py

@@ -117,7 +117,17 @@ def sample_ome_tiff(remote_sample: Callable) -> Path:
     Download ome-tiff image for pytest.
 
     """
-    return remote_sample("ome-brightfield-pyramid-1-small")
+    return remote_sample("ome-brightfield-small-pyramid")
+
+
+@pytest.fixture(scope="session")
+def sample_ome_tiff_level_0(remote_sample: Callable) -> Path:
+    """Sample pytest fixture for ome-tiff image with one level.
+
+    Download ome-tiff image for pytest.
+
+    """
+    return remote_sample("ome-brightfield-small-level-0")
 
 
 @pytest.fixture(scope="session")

tests/test_tiffreader.py

@@ -13,7 +13,7 @@ def test_ome_missing_instrument_ref(
     remote_sample: Callable,
 ) -> None:
     """Test that an OME-TIFF can be read without instrument reference."""
-    sample = remote_sample("ome-brightfield-pyramid-1-small")
+    sample = remote_sample("ome-brightfield-small-level-0")
     wsi = wsireader.TIFFWSIReader(sample)
     page = wsi.tiff.pages[0]
     description = page.description
@@ -37,7 +37,7 @@ def test_ome_missing_physicalsize(
     remote_sample: Callable,
 ) -> None:
     """Test that an OME-TIFF can be read without physical size."""
-    sample = remote_sample("ome-brightfield-pyramid-1-small")
+    sample = remote_sample("ome-brightfield-small-level-0")
     wsi = wsireader.TIFFWSIReader(sample)
     page = wsi.tiff.pages[0]
     description = page.description
@@ -62,7 +62,7 @@ def test_ome_missing_physicalsizey(
     remote_sample: Callable,
 ) -> None:
     """Test that an OME-TIFF can be read without physical size."""
-    sample = remote_sample("ome-brightfield-pyramid-1-small")
+    sample = remote_sample("ome-brightfield-small-level-0")
     wsi = wsireader.TIFFWSIReader(sample)
     page = wsi.tiff.pages[0]
     description = page.description
@@ -86,7 +86,7 @@ def test_tiffreader_non_tiled_metadata(
     remote_sample: Callable,
 ) -> None:
     """Test that fetching metadata for non-tiled TIFF works."""
-    sample = remote_sample("ome-brightfield-pyramid-1-small")
+    sample = remote_sample("ome-brightfield-small-level-0")
     wsi = wsireader.TIFFWSIReader(sample)
     monkeypatch.setattr(wsi.tiff, "is_ome", False)
     monkeypatch.setattr(

tests/test_wsireader.py

@@ -959,15 +959,23 @@ def test_read_bounds_interpolated(sample_svs: Path) -> None:
 
 
 def test_read_bounds_level_consistency_openslide(sample_ndpi: Path) -> None:
-    """Test read_bounds produces the same visual field across resolution levels."""
+    """Test read_bounds produces the same visual field across resolution levels.
+
+    with OpenSlideWSIReader.
+
+    """
     wsi = wsireader.OpenSlideWSIReader(sample_ndpi)
     bounds = NDPI_TEST_TISSUE_BOUNDS
 
     read_bounds_level_consistency(wsi, bounds)
 
 
 def test_read_bounds_level_consistency_jp2(sample_jp2: Path) -> None:
-    """Test read_bounds produces the same visual field across resolution levels."""
+    """Test read_bounds produces the same visual field across resolution levels.
+
+    Using JP2WSIReader.
+
+    """
     bounds = JP2_TEST_TISSUE_BOUNDS
     wsi = wsireader.JP2WSIReader(sample_jp2)
 
@@ -1883,11 +1891,11 @@ def test_tiffwsireader_invalid_svs_metadata(
 
 
 def test_tiffwsireader_invalid_ome_metadata(
-    sample_ome_tiff: Path,
+    sample_ome_tiff_level_0: Path,
     monkeypatch: pytest.MonkeyPatch,
 ) -> None:
     """Test exception raised for invalid OME-XML metadata instrument."""
-    wsi = wsireader.TIFFWSIReader(sample_ome_tiff)
+    wsi = wsireader.TIFFWSIReader(sample_ome_tiff_level_0)
     monkeypatch.setattr(
         wsi.tiff.pages[0],
         "description",
@@ -2545,7 +2553,7 @@ def test_jp2_no_header(tmp_path: Path, monkeypatch: pytest.MonkeyPatch) -> None:
         },
         {
             "reader_class": TIFFWSIReader,
-            "sample_key": "ome-brightfield-pyramid-1-small",
+            "sample_key": "ome-brightfield-small-pyramid",
             "kwargs": {},
         },
         {

tiatoolbox/data/remote_samples.yaml

@@ -25,8 +25,10 @@ files:
     url: [*wsis, "CMU-1-Small-Region.jpeg.tiff"]
   tiled-tiff-1-small-jp2k:
     url: [*wsis, "CMU-1-Small-Region.jp2k.tiff"]
-  ome-brightfield-pyramid-1-small:
-    url: [*wsis, "CMU-1-Small-Region.ome.tiff"]
+  ome-brightfield-small-level-0:
+    url: [*wsis, "CMU-1-Small-Region-Level-0.ome.tiff"]
+  ome-brightfield-small-pyramid:
+    url: [*wsis, "CMU-1-Small-Region-Pyramid.ome.tif"]
   two-tiled-pages:
     url: [*wsis, "two-tiled-pages.tiff"]
   ventana-tif:

tiatoolbox/visualization/bokeh_app/main.py

@@ -2086,7 +2086,7 @@ def setup_doc(self: DocConfig, base_doc: Document) -> tuple[Row, Tabs]:
 
         # Set initial slide to first one in base folder
         slide_list = []
-        for ext in ["*.svs", "*ndpi", "*.tiff", "*.mrxs", "*.png", "*.jpg"]:
+        for ext in ["*.svs", "*ndpi", "*.tiff", "*.tif", "*.mrxs", "*.png", "*.jpg"]:
             slide_list.extend(list(doc_config["slide_folder"].glob(ext)))
             slide_list.extend(
                 list(doc_config["slide_folder"].glob(str(Path("*") / ext))),

tiatoolbox/wsicore/wsireader.py

@@ -2,7 +2,6 @@
 
 from __future__ import annotations
 
-import copy
 import json
 import logging
 import math
@@ -380,7 +379,9 @@ def open(  # noqa: PLR0911
                 )
             return NGFFWSIReader(input_path, mpp=mpp, power=power)
 
-        if suffixes[-2:] in ([".ome", ".tiff"],):
+        if suffixes[-2:] in ([".ome", ".tiff"],) or suffixes[-2:] in (
+            [".ome", ".tif"],
+        ):
             return TIFFWSIReader(input_path, mpp=mpp, power=power)
 
         if last_suffix in (".tif", ".tiff"):
@@ -477,7 +478,7 @@ def info(self: WSIReader) -> WSIMeta:
 
         """
         if self._m_info is not None:
-            return copy.deepcopy(self._m_info)
+            return self._m_info
         self._m_info = self._info()
         if self._manual_mpp:
             self._m_info.mpp = np.array(self._manual_mpp)
@@ -3472,7 +3473,14 @@ def __init__(
                 len(self.tiff.pages) == 1,
             ],
         )
-        if not any([self.tiff.is_svs, self.tiff.is_ome, is_single_page_tiled]):
+        if not any(
+            [
+                self.tiff.is_svs,
+                self.tiff.is_ome,
+                is_single_page_tiled,
+                self.tiff.is_bigtiff,
+            ]
+        ):
             msg = "Unsupported TIFF WSI format."
             raise ValueError(msg)
 
@@ -3503,9 +3511,16 @@ def page_area(page: tifffile.TiffPage) -> float:
             group[0] = self._zarr_group
             self._zarr_group = group
         self.level_arrays = {
-            int(key): ArrayView(array, axes=self.info.axes)
+            int(key): ArrayView(array, axes=self._axes)
             for key, array in self._zarr_group.items()
         }
+        # ensure level arrays are sorted by descending area
+        self.level_arrays = dict(
+            sorted(
+                self.level_arrays.items(),
+                key=lambda x: -np.prod(self._canonical_shape(x[1].array.shape[:2])),
+            )
+        )
 
     def _canonical_shape(self: TIFFWSIReader, shape: IntPair) -> tuple:
         """Make a level shape tuple in YXS order.
@@ -3761,10 +3776,10 @@ def _info(self: TIFFWSIReader) -> WSIMeta:
                 Containing metadata.
 
         """
-        level_count = len(self._zarr_group)
+        level_count = len(self.level_arrays)
         level_dimensions = [
-            np.array(self._canonical_shape(p.shape)[:2][::-1])
-            for p in self._zarr_group.values()
+            np.array(self._canonical_shape(p.array.shape)[:2][::-1])
+            for p in self.level_arrays.values()
         ]
         slide_dimensions = level_dimensions[0]
         level_downsamples = [(level_dimensions[0] / x)[0] for x in level_dimensions]
@@ -4007,10 +4022,10 @@ def read_rect(
         # Find parameters for optimal read
         (
             read_level,
-            _,
-            _,
+            level_read_location,
+            level_read_size,
             post_read_scale,
-            baseline_read_size,
+            _,
         ) = self.find_read_rect_params(
             location=location,
             size=size,
@@ -4019,16 +4034,15 @@ def read_rect(
         )
 
         bounds = utils.transforms.locsize2bounds(
-            location=location,
-            size=baseline_read_size,
+            location=level_read_location,
+            size=level_read_size,
         )
         im_region = utils.image.safe_padded_read(
             image=self.level_arrays[read_level],
             bounds=bounds,
             pad_mode=pad_mode,
             pad_constant_values=pad_constant_values,
         )
-
         im_region = utils.transforms.imresize(
             img=im_region,
             scale_factor=post_read_scale,
@@ -4163,7 +4177,7 @@ class docstrings for more information.
             # but base image is of different scale)
             (
                 read_level,
-                _,
+                bounds_at_read_level,
                 _,
                 post_read_scale,
             ) = self._find_read_bounds_params(
@@ -4175,7 +4189,7 @@ class docstrings for more information.
             # Find parameters for optimal read
             (
                 read_level,
-                _,
+                bounds_at_read_level,
                 size_at_requested,
                 post_read_scale,
             ) = self._find_read_bounds_params(
@@ -4186,7 +4200,7 @@ class docstrings for more information.
 
         im_region = utils.image.sub_pixel_read(
             image=self.level_arrays[read_level],
-            bounds=bounds_at_baseline,
+            bounds=bounds_at_read_level,
             output_size=size_at_requested,
             interpolation=interpolation,
             pad_mode=pad_mode,