Migrate from AICSImage to BioImage

.github/lint/action.yml

@@ -10,7 +10,7 @@ runs:
   using: "composite"
   steps:
     - run: |
-        pip install fawltydeps==0.12.0
+        pip install fawltydeps
         fawltydeps --check --detailed --verbose
       shell: bash -l {0}
       working-directory: ${{ inputs.directory }}

core/lls_core/deconvolution.py

@@ -7,9 +7,9 @@
 import logging
 import importlib.util
 from typing import Collection, Iterable,Union,Literal, Optional, TYPE_CHECKING
-from aicsimageio.aics_image import AICSImage
+from bioio import BioImage
+import bioio_czi
 from skimage.io import imread
-from aicspylibczi import CziFile
 from numpy.typing import NDArray
 import os
 import numpy as np
@@ -62,12 +62,10 @@ def read_psf(psf_paths: Collection[Path],
     for psf in psf_paths:
         if psf.exists() and psf.is_file():
             if psf.suffix == ".czi":
-                psf_czi = CziFile(psf.__str__())
-                psf_aics = psf_czi.read_image()
+                psf_czi = BioImage(psf.__str__(), reader=bioio_czi.Reader)
+                psf_aics = psf_czi.data
                 # make sure shape is 3D
-                psf_aics = psf_aics[0][0]  # np.expand_dims(psf_aics[0],axis=0)
-                # if len(psf_aics[0])>=1:
-                #psf_channels = len(psf_aics[0])
+                psf_aics = psf_aics[0][0]  
                 assert len(
                     psf_aics.shape) == 3, f"PSF should be a 3D image (shape of 3), but got {psf_aics.shape}"
                 # pad psf to multiple of 16 for decon
@@ -79,8 +77,8 @@ def read_psf(psf_paths: Collection[Path],
                    if len(psf_aics_data.shape) != 3:
                        raise ValueError(f"PSF should be a 3D image (shape of 3), but got {psf_aics.shape}")
                 else:
-                    #Use AICSImageIO
-                    psf_aics = AICSImage(str(psf))
+                    #Use BioIO
+                    psf_aics = BioImage(str(psf))
                     psf_aics_data = psf_aics.data[0][0]
                     psf_aics_data = pad_image_nearest_multiple(
                         img=psf_aics_data, nearest_multiple=16)           

core/lls_core/models/deskew.py

@@ -16,7 +16,7 @@
 import numpy as np
 
 if TYPE_CHECKING:
-    from aicsimageio.types import PhysicalPixelSizes
+    from bioio import PhysicalPixelSizes
 
 class DefinedPixelSizes(FieldAccessModel):
     """
@@ -161,7 +161,7 @@ def convert_skew(cls, v: Any):
 
     @validator("physical_pixel_sizes", pre=True, always=True)
     def convert_pixels(cls, v: Any, values: dict[Any, Any]):
-        from aicsimageio.types import PhysicalPixelSizes
+        from bioio import PhysicalPixelSizes
         if isinstance(v, PhysicalPixelSizes):
             v = DefinedPixelSizes.from_physical(v)
         elif isinstance(v, tuple) and len(v) == 3:
@@ -176,15 +176,15 @@ def convert_pixels(cls, v: Any, values: dict[Any, Any]):
 
     @root_validator(pre=True)
     def read_image(cls, values: dict):
-        from aicsimageio import AICSImage
+        from bioio import BioImage
         from os import fspath
 
         img = values["input_image"]
 
-        aics: AICSImage | None = None
+        aics: BioImage | None = None
         if is_pathlike(img):
-            aics = AICSImage(fspath(img))
-        elif isinstance(img, AICSImage):
+            aics = BioImage(fspath(img))
+        elif isinstance(img, BioImage):
             aics = img
         elif isinstance(img, DataArray):
             if set(img.dims) >= {"Z", "Y", "X"}:
@@ -198,9 +198,9 @@ def read_image(cls, values: dict):
             else:
                 raise ValueError("Only 3D numpy arrays are currently supported. If you have a different shape, please use a DataArray and name your dimensions C, T, Z, Y and/or Z.")
         else:
-            raise ValueError("Value of input_image was neither a path, an AICSImage, or array-like.")
+            raise ValueError("Value of input_image was neither a path, a BioImage, or array-like.")
 
-        # If the image was convertible to AICSImage, we should use the metadata from there
+        # If the image was convertible to BioImage, we should use the metadata from there
         if aics:
             values["input_image"] = aics.xarray_dask_data 
             # Take pixel sizes from the image metadata, but only if they're defined

core/lls_core/types.py

@@ -6,7 +6,7 @@
 import numpy as np
 from numpy.typing import NDArray
 from xarray import DataArray
-from aicsimageio import AICSImage
+from bioio import BioImage
 from os import fspath, PathLike as OriginalPathLike
 
 # This is a superset of os.PathLike
@@ -19,20 +19,20 @@ def is_pathlike(x: Any) -> TypeGuard[PathLike]:
 def is_arraylike(arr: Any) -> TypeGuard[ArrayLike]:
     return isinstance(arr, (DaskArray, np.ndarray, OCLArray, DataArray))
 
-ImageLike: TypeAlias = Union[PathLike, AICSImage, ArrayLike]
-def image_like_to_image(img: ImageLike) -> DataArray:
-    """
-    Converts an image in one of many formats to a DataArray
-    """
-    # First try treating it as a path
-    try:
-        img = AICSImage(fspath(img))
-    except TypeError:
-        pass
-    if isinstance(img, AICSImage):
-        return img.xarray_dask_data
-    else:
-        for required_key in ("shape", "dtype", "ndim", "__array__", "__array_ufunc__"):
-            if not hasattr(img, required_key):
-                raise ValueError(f"The provided object {img} is not array like!")
+ImageLike: TypeAlias = Union[PathLike, BioImage, ArrayLike]
+def image_like_to_image(img: ImageLike) -> DataArray:
+    """
+    Converts an image in one of many formats to a DataArray
+    """
+    # First try treating it as a path
+    try:
+        img = BioImage(fspath(img))
+    except TypeError:
+        pass
+    if isinstance(img, BioImage):
+        return img.xarray_dask_data
+    else:
+        for required_key in ("shape", "dtype", "ndim", "__array__", "__array_ufunc__"):
+            if not hasattr(img, required_key):
+                raise ValueError(f"The provided object {img} is not array like!")
         return DataArray(img)

core/lls_core/writers.py

@@ -118,7 +118,8 @@ def flush(self):
                 str(path),
                 data = images_array,
                 bigtiff=True,
-                resolution=(1./self.lattice.dx, 1./self.lattice.dy, "MICROMETER"),
+                resolution=(1./self.lattice.dx, 1./self.lattice.dy),
+                resolutionunit="MICROMETER",
                 metadata={'spacing': self.lattice.new_dz, 'unit': 'um', 'axes': 'TZCYX'},
                 imagej=True
             )

core/pyproject.toml

@@ -36,10 +36,11 @@ classifiers = [
 ]
 requires-python = ">=3.8"
 dependencies = [
-    "aicsimageio>=4.6.3",
-    # Earlier versions don't have Python 3.11 binaries, and the sdist
-    # is misconfigured: https://github.com/AllenCellModeling/aicspylibczi/issues/90
-    "aicspylibczi>=3.1.1",
+    "bioio>=3.0.0",
+    "bioio-bioformats>=1.3.0",
+    "bioio-czi>=2.4.0",
+    "bioio-ome-tiff>=1.4.0",
+    "bioio-tifffile==1.3.0",
     "click",
     "dask",
     "dask[distributed]",
@@ -158,7 +159,12 @@ ignore_unused = [
 
     # Used for the deployment, but never imported
     "build",
-    "twine"
+    "twine",
+
+    #bioio packages are needed but not directly imported
+    "bioio-bioformats",
+    "bioio-czi",
+    "bioio-ome-tiff"
 ]
 output_format = "human_detailed"  
 

core/tests/test_cli.py

@@ -2,7 +2,7 @@
 
 from typing import Callable, List
 import pytest
-from aicsimageio.aics_image import AICSImage
+from bioio import BioImage
 from npy2bdv import BdvEditor
 import numpy as np
 from pathlib import Path
@@ -14,8 +14,9 @@ def create_image(path: Path):
     # Create a zero array of shape 5x5x5 with a value of 10 at (2,4,2)
     raw = np.zeros((5, 5, 5))
     raw[2, 4, 2] = 10
-    # Save image as a tif filw in home directory
-    AICSImage(raw).save(path)
+    # Save image as a tif file in home directory
+    b = BioImage(raw, physical_pixel_sizes={ax: 1. for ax in 'ZYX'})
+    b.save(path)
     assert path.exists()
 
 
@@ -27,8 +28,9 @@ def create_data(dir: Path) -> Path:
     # Create a zero array of shape 5x5x5 with a value of 10 at (2,4,2)
     raw = np.zeros((5, 5, 5))
     raw[2, 4, 2] = 10
-    # Save image as a tif filw in home directory
-    AICSImage(raw).save(input_file)
+    # Save image as a tif file in home directory
+    b = BioImage(raw, physical_pixel_sizes={ax: 1. for ax in 'ZYX'})
+    b.save(input_file)
     assert input_file.exists()
 
     config: dict[str, str] = {
@@ -47,7 +49,7 @@ def assert_tiff(output_dir: Path):
     results = list(output_dir.glob("*.tif"))
     assert len(results) > 0
     for result in results:
-        AICSImage(result).get_image_data()
+        BioImage(result).get_image_data()
 
 def assert_h5(output_dir: Path):
     """Checks that a valid H5 was generated"""

core/tests/test_process.py

@@ -9,6 +9,8 @@
 from pathlib import Path
 from napari_workflows import Workflow
 from pytest import FixtureRequest
+from bioio_tifffile import reader as tiffreader
+from bioio import BioImage
 
 
 from .params import parameterized
@@ -51,22 +53,30 @@ def test_save(minimal_image_path: str, args: dict):
 
 
 def test_process_deconv_crop():
-    for slice in (
-        LatticeData.parse_obj(
-            {
-                "input_image": root / "raw.tif",
-                "deconvolution": {
-                    "psf": [root / "psf.tif"],
-                },
-                "crop": CropParams(
-                    roi_list=[[[0, 0], [0, 110], [95, 0], [95, 110]]]
-                ),
-            }
-        )
-        .process()
-        .slices
-    ):
-        assert slice.data.ndim == 3
+    with tempfile.TemporaryDirectory() as tempdir:
+        for slice in (
+            LatticeData.parse_obj(
+                {
+                    # use BioImage rather than just Path to ensure tifffile
+                    # is used instead of bioformats, which prevents dask use
+                    # due to a bug:
+                    # https://github.com/bioio-devs/bioio-bioformats/issues/40
+                    "input_image": BioImage(
+                        root / "raw.tif", reader=tiffreader.Reader
+                    ),
+                    "deconvolution": {
+                        "psf": [root / "psf.tif"],
+                    },
+                    "crop": CropParams(
+                        roi_list=[[[0, 0], [0, 110], [95, 0], [95, 110]]]
+                    ),
+                    "save_dir": tempdir,
+                }
+            )
+            .process()
+            .slices
+        ):
+            assert slice.data.ndim == 3
 
 
 def test_process_time_range(multi_channel_time: Path):

core/tests/test_validation.py

@@ -48,7 +48,7 @@ def test_allow_trailing_slash():
 
 def test_infer_czi_pixel_sizes(rbc_tiny: Path):
     mock = PropertyMock()
-    with patch("aicsimageio.AICSImage.physical_pixel_sizes", new=mock):
+    with patch("bioio.BioImage.physical_pixel_sizes", new=mock):
         DeskewParams(input_image=rbc_tiny)
-        # The AICSImage should be queried for the pixel sizes
+        # The BioImage should be queried for the pixel sizes
         assert mock.called

core/tests/utils.py

@@ -3,7 +3,7 @@
 from typer.testing import CliRunner
 from lls_core.cmds.__main__ import app
 import npy2bdv
-from aicsimageio import AICSImage
+from bioio import BioImage
 
 def invoke(args: Sequence[str]):
     CliRunner().invoke(app, args, catch_exceptions=False)
@@ -13,5 +13,5 @@ def valid_image_path(path: Path) -> bool:
         npy2bdv.npy2bdv.BdvEditor(str(path)).read_view()
         return True
     else:
-        AICSImage(path).get_image_data()
+        BioImage(path).get_image_data()
         return True

plugin/napari_lattice/fields.py

@@ -402,7 +402,7 @@ def _get_kwargs(self) -> DeskewKwargs:
         """
         Returns the LatticeData fields that the Deskew tab can provide
         """
-        from aicsimageio.types import PhysicalPixelSizes
+        from bioio import PhysicalPixelSizes
         DeskewParams.update_forward_refs()
         params = lattice_params_from_napari(
                 imgs=self.img_layer.value,

plugin/napari_lattice/reader.py

@@ -13,18 +13,18 @@
 import os 
 import numpy as np
 from napari.layers import Image
-from aicsimageio.aics_image import AICSImage
+from bioio import BioImage
 
 from typing import List, Optional, Tuple, Collection, TYPE_CHECKING, TypedDict
 
-from aicsimageio.types import PhysicalPixelSizes
+from bioio import PhysicalPixelSizes
 from lls_core.models.deskew import DefinedPixelSizes
 
 from logging import getLogger
 logger = getLogger(__name__)
 
 if TYPE_CHECKING:
-    from aicsimageio.types import ImageLike
+    from bioio import ImageLike
     from xarray import DataArray
 
 class NapariImageParams(TypedDict):
@@ -74,7 +74,7 @@ def lattice_params_from_napari(
         save_names.append(save_name)
 
         if 'aicsimage' in img.metadata.keys():
-            img_data_aics: AICSImage = img.metadata['aicsimage']
+            img_data_aics: BioImage = img.metadata['aicsimage']
             # If the user has not provided pixel sizes, we extract them fro the metadata
             # Only process pixel sizes that are not none
             if physical_pixel_sizes is None and all(img_data_aics.physical_pixel_sizes):
@@ -191,13 +191,13 @@ def bdv_h5_reader(path):
     layer_type = "image"  # optional, default is "image"
     return [(images, add_kwargs, layer_type)]
 
-def tiff_reader(path: ImageLike) -> List[Tuple[AICSImage, dict, str]]:
+def tiff_reader(path: ImageLike) -> List[Tuple[BioImage, dict, str]]:
     """Take path to tiff image and returns a list of LayerData tuples.
-    Specifying tiff_reader to have better control over tifffile related errors when using AICSImage
+    Specifying tiff_reader to have better control over tifffile related errors when using BioImage
     """
 
     try:
-        image = AICSImage(path)
+        image = BioImage(path)
     except Exception as e:
         raise Exception("Error reading TIFF. Try upgrading tifffile library: pip install tifffile --upgrade.") from e
 

plugin/pyproject.toml

@@ -36,7 +36,11 @@ classifiers = [
 ]
 requires-python = ">=3.8"
 dependencies = [
-    "aicsimageio>=4.6.3",
+    "bioio>=3.0.0",
+    "bioio-bioformats>=1.3.0",
+    "bioio-czi>=2.4.0",
+    "bioio-ome-tiff>=1.4.0",
+    "bioio-tifffile==1.3.0",
     "dask[distributed]",
     # This isn't used directly, but we need to pin this version
     "fsspec>=2022.8.2",
@@ -45,7 +49,6 @@ dependencies = [
     # The lower bound is because we need this Python 3.8 fix: https://github.com/hanjinliu/magic-class/pull/108
     "magic-class>=0.7.5",
     "magicgui<0.8.0",
-    "napari-aicsimageio>=0.7.2",
     "napari-workflow-inspector",
     "napari-workflows>=0.2.8",
     "napari[all]>=0.4.11",
@@ -99,7 +102,12 @@ ignore_unused = [
     # This is pinned but unused
     "fsspec",
     "imageio",
-    "ome-types"
+    "ome-types",
+
+    #bioio packages are needed but not directly imported
+    "bioio-bioformats",
+    "bioio-czi",
+    "bioio-ome-tiff"
 ]
 output_format = "human_detailed"