tiatoolbox 1.3.0

Computational pathology toolbox developed by TIA Centre.

TIA Toolbox

Computational Pathology Toolbox developed at the TIA Centre

Getting Started

All Users

This package is for those interested in digital pathology: including graduate students, medical staff, members of the TIA Centre and of PathLAKE, and anyone, anywhere, who may find it useful. We will continue to improve this package, taking account of developments in pathology, microscopy, computing and related disciplines. Please send comments and criticisms to [e-mail].

tiatoolbox is a multipurpose name that we use for 1) a certain computer program, 2) a Python package of related programs, created by us at the TIA Centre to help people get started in Digital Pathology, 3) this repository, 4) a certain virtual environment.

Developers

Anyone wanting to contribute to this repository, please first look at our Wiki and at our web page for contributors. See also the Prepare for development section of this document.

Links, if needed

The bash shell is available on all commonly encountered platforms. Commands in this README are in bash. Windows users can use the command prompt to install conda and python packages.

conda is a management system for software packages and virtual environments. To get conda, download Anaconda, which includes hundreds of the most useful Python packages, using 2GB disk space. Alternatively, miniconda uses 400MB, and packages can be added as needed.

GitHub is powered by the version control system git, which has many users and uses. In GitHub, it is used to track versions of code and other documents.

Examples Taster

1. Click here for jupyter notebooks, hosted on the web, with demos of tiatoolbox. All necessary resources to run the notebooks are remotely provided, so you don't need to have Python installed on your computer.
2. Click on a filename with suffix .ipynb and the notebook will open in your browser.
3. Click on one of the two blue checkboxes in your browser window labelled either Open in Colab or Open in Kaggle: colab and kaggle are websites providing free-of-charge platforms for running jupyter notebooks.
4. Operate the notebook in your browser, editing, inserting or deleting cells as desired.
5. Changes you make to the notebook will last no longer than your colab or kaggle session.

Install Python package

If you wish to use our programs, perhaps without developing them further, run the command pip install tiatoolbox or pip install --ignore-installed --upgrade tiatoolbox to upgrade from an existing installation. Detailed installation instructions can be found in the documentation.

To understand better how the programs work, study the jupyter notebooks referred to under the heading Examples Taster.

Command Line

tiatoolbox supports various features through command line. For more information, please try tiatoolbox --help

Prepare for development

Prepare a computer as a convenient platform for further development of the Python package tiatoolbox and related programs as follows.

1. Install pre-requisite software
2. Open a terminal window
   

    $ cd <future-home-of-tiatoolbox-directory>

3. Download a complete copy of the tiatoolbox.

    $ git clone https://github.com/TissueImageAnalytics/tiatoolbox.git

4. Change directory to tiatoolbox

    $ cd tiatoolbox

5. Create virtual environment for TIAToolbox using

    $ conda env create -f requirements.dev.conda.yml # for linux/mac only.
    $ conda activate tiatoolbox-dev

or

    $ conda create -n tiatoolbox-dev python=3.8 # select version of your choice
    $ conda activate tiatoolbox-dev
    $ pip install -r requirements_dev.txt

6. To use the packages installed in the environment, run the command:

    $ conda activate tiatoolbox-dev

License

The source code TIA Toolbox (tiatoolbox) as hosted on GitHub is released under the The 3-Clause BSD License.

The full text of the licence is included in LICENSE.

Cite this repository

If you find TIAToolbox useful or use it in your research, please consider citing our paper:

@article{
    Pocock2022,
    author = {Pocock, Johnathan and Graham, Simon and Vu, Quoc Dang and Jahanifar, Mostafa and Deshpande, Srijay and Hadjigeorghiou, Giorgos and Shephard, Adam and Bashir, Raja Muhammad Saad and Bilal, Mohsin and Lu, Wenqi and Epstein, David and Minhas, Fayyaz and Rajpoot, Nasir M and Raza, Shan E Ahmed},
    doi = {10.1038/s43856-022-00186-5},
    issn = {2730-664X},
    journal = {Communications Medicine},
    month = {sep},
    number = {1},
    pages = {120},
    publisher = {Springer US},
    title = {{TIAToolbox as an end-to-end library for advanced tissue image analytics}},
    url = {https://www.nature.com/articles/s43856-022-00186-5},
    volume = {2},
    year = {2022}
}

Auxiliary Files

Auxiliary files, such as pre-trained model weights downloaded from the TIA Centre webpage (https://warwick.ac.uk/tia/), are provided under the Creative Commons Attribution-NonCommercial-ShareAlike Version 4 (CC BY-NC-SA 4.0) license.

Dual License

If you would like to use any of the source code or auxiliary files (e.g. pre-trained model weights) under a different license agreement please contact the Tissue Image Analytics (TIA) Centre at the University of Warwick ([e-mail]).

History

1.3.0 (2022-10-20)

Major Updates and Feature Improvements

• Adds an AnnotationTileGenerator and AnnotationRenderer which allows serving of tiles rendered directly from an annotation store.
• Adds DFBR registration model and jupyter notebook example
  • Adds DICE metric
• Adds SCCNN architecture. [read the docs]
• Adds MapDe architecture. [read the docs]
• Adds support for reading MPP metadata from NGFF v0.4
• Adds enhancements to tiatoolbox.annotation.storage that are useful when using an AnnotationStore for visualization purposes.

Changes to API

• None

Bug Fixes and Other Changes

• Fixes colorbar_params #410
• Fixes Jupyter notebooks for better read the docs rendering
  • Fixes typos, metadata and links
• Fixes nucleus_segmentor_engine for boundary artefacts
• Fixes the colorbar cropping in tests
• Adds citation in README.md and CITATION.cff to Nature Communications Medicine paper
• Fixes a bug #452 raised by @rogertrullo where only the numerator of the TIFF resolution tags was being read.
• Fixes HoVer-Net+ post-processing to be inline with original work.
• Fixes a bug where an exception would be raised if the OME XML is missing objective power.

Development related changes

• Uses Furo theme for readthedocs
• Replaces nbgallery and nbsphinx with myst-nb for jupyter notebook rendering
• Uses myst for markdown parsing
• Uses requirements.txt to define dependencies for requirements consistency
• Adds notebook AST pre-commit hook
• Adds check to validate python examples in the code
• Adds check to resolve imports
• Fixes an error in a docstring which triggered the failing test.
• Adds pre-commit hooks to format markdown and notebook markdown
• Adds pip install workflow to resolve dependencies when requirements file is updated
• Improves tiatoolbox import using LazyLoader

1.2.1 (2022-07-07)

Major Updates and Feature Improvements

• None

Changes to API

• None

Bug Fixes and Other Changes

• Fixes issues with dependencies.
  • Adds flask to dependencies.
• Fixes missing file in the python package.
• Clarifies help string for show-wsi option.

Development related changes

• Removes Travis CI.
  • GitHub Actions will be used instead.
• Adds pre-commit hooks to check requirements consistency.
• Adds GitHub Action to resolve conda environment checks on Windows and Ubuntu.

1.2.0 (2022-07-05)

Major Updates and Feature Improvements

• Adds support for Python 3.10
• Adds short description for IDARS algorithm #383
• Adds support for NGFF v0.4 OME-ZARR.
• Adds CLI for launching tile server.

Changes to API

• Renames stainnorm_target() function to stain_norm_target().
• Removes get_wsireader
• Replaces the custom PlattScaler in tools/scale.py with the regular Scikit-Learn LogisticRegression.

Bug Fixes and Other Changes

• Fixes bugs in UNET architecture.
  • Number of channels in Batchnorm argument in the decoding path to match with the input channels.
  • Padding 0 creates feature maps in the decoder part with the same size as encoder.
• Fixes linter issues and typos
• Fixes incorrect output with overlap in predictor.merge_predictions() and return_raw=True
  • Thanks to @paulhacosta for raising #356, Fixed by #358.
• Fixes errors with JP2 read. Checks input path exists.
• Fixes errors with torch upgrade to 1.12.

Development related changes

• Adds pre-commit hooks for consistency across the repo.
• Sets up GitHub Actions Workflow.
  • Travis CI will be removed in future release.

1.1.0 (2022-05-07)

Major Updates and Feature Improvements

• Adds DICOM Support.
• Updates license to more permissive BSD 3-clause.
• Adds micronet model.
• Improves support for tiff files.
  • Adds a check for tiles in a TIFF file when opening.
  • Uses OpenSlide to read a TIFF if it has tiles instead of OpenCV (VirtualWSIReader).
  • Adds a fallback to tifffile if it is tiled but openslide cannot read it (e.g. jp2k or jpegxl tiles).
• Adds support for multi-channel images (HxWxC).
• Fixes performance issues in semantic_segmentor.py.
  • Performance gain measurement: 21.67s (new) vs 45.564 (old) using a 4k x 4k WSI.
  • External Contribution from @ByteHexler.
• Adds benchmark for Annotations Store.

Changes to API

• None

Bug Fixes and Other Changes

• Enhances the error messages to be more informative.
• Fixes Flake8 Errors, typos.
  • Fixes patch predictor models based after fixing a typo.
• Bug fixes in Graph functions.
• Adds documentation for docker support.
• General tidying up of docstrings.
• Adds metrics to readthedocs/docstrings for pretrained models.

Development related changes

• Adds pydicom and wsidicom as dependency.
• Updates dependencies.
• Fixes Travis detection and makes improvements to run tests faster on Travis.
• Adds Dependabot to automatically update dependencies.
• Improves CLI definitions to make it easier to integrate new functions.
• Fixes compile options for test_annotation_stores.py

1.0.1 (2022-01-31)

Major Updates and Feature Improvements

• Updates dependencies for conda recipe #262

Changes to API

• None

Bug Fixes and Other Changes

• Adds User Warning For Missing SQLite Functions
• Fixes Pixman version check errors
• Fixes empty query in instance segmentor

Development related changes

• Fixes flake8 linting issues and typos
• Conditional pytest.skipif to skip GPU tests on travis while running them locally or elsewhere

1.0.0 (2021-12-23)

Major Updates and Feature Improvements

• Adds nucleus instance segmentation base class
  • Adds HoVerNet architecture
• Adds multi-task segmentor HoVerNet+ model
• Adds IDaRS pipeline
• Adds SlideGraph pipeline
• Adds PCam patch classification models
• Adds support for stain augmentation feature
• Adds classes and functions under tiatoolbox.tools.graph to enable construction of graphs in a format which can be used with PyG (PyTorch Geometric).
• Add classes which act as a mutable mapping (dictionary like) structure and enables efficient management of annotations. (#135)
• Adds example notebook for adding advanced models
• Adds classes which can generate zoomify tiles from a WSIReader object.
• Adds WSI viewer using Zoomify/WSIReader API (#212)
• Adds README to example page for clarity
• Adds support to override or specify mpp and power

Changes to API

• Replaces models.controller API with models.engine
• Replaces CNNPatchPredictor with PatchPredictor

Bug Fixes and Other Changes

• Fixes Fix filter_coordinates read wrong resolutions for patch extraction
• For PatchPredictor
  • ioconfig will supersede everything
  • if ioconfig is not provided
    • If model is pretrained (defined in pretrained_model.yaml )
      • Use the yaml ioconfig
      • Any other input patch reading arguments will overwrite the yaml ioconfig (at the same keyword).
    • If model is not defined, all input patch reading arguments must be provided else exception will be thrown.
• Improves performance of mask based patch extraction

Development related changes

• Improve tests performance for Travis runs
• Adds feature detection mechanism to detect the platform and installed packages etc.
• On demand imports for some libraries for performance
• Improves performance of mask based patch extraction

0.8.0 (2021-10-27)

Major Updates and Feature Improvements

• Adds SemanticSegmentor which is Predictor equivalent for semantic segmentation.
• Add TIFFWSIReader class to support OMETiff reading.
• Adds FeatureExtractor API to controller.
• Adds WSI Serialization Dataset which support changing parallel workers on the fly. This would reduce the time spent to create new worker for every WSI/Tile (costly).
• Adds IOState data class to contain IO information for loading input to model and assembling model output back to WSI/Tile.
• Minor updates for get_coordinates to pave the way for getting patch IO for segmentation.
• Migrates old code to new variable names (patch extraction, patch wsi model).
• Change in API from pretrained_weight to pretrained_weights.
• Adds cli for semantic segmentation.
• Update python notebooks to add read_rect and read_bounds examples with mpp read.

Changes to API

• Adds WSIReader.open. get_wsireader will deprecate in the next release. Please use WSIReader.open instead.
• CLI is now POSIX compatible
  • Replaces underscores in variable names with hyphens
• Models API updated to use pretrained_weights instead of pretrained_weight.
• Move string_to_tuple to tiatoolbox/utils/misc.py

Bug Fixes and Other Changes

• Fixes README git clone instructions.
• Fixes stain normalisation due to changes in sklearn.
• Fixes a test in tests/test_slide_info
• Fixes readthedocs documentation issues

Development related changes

• Adds dependencies for tiffile, imagecodecs, zarr.
• Adds more stringent pre-commit checks
• Moved local test files into tiatoolbox/data.
• Fixed Manifest.ini and added tiatoolbox/data. This means that this directory will be downloaded with the package.
• Using pkg_resources to properly load bundled resources (e.g. target_image.png) in tiatoolbox.data.
• Removed duplicate code in conftest.py for downloading remote files. This is now in tiatoolbox.data._fetch_remote_file.
• Fixes errors raised by new flake8 rules.
  • Remove leading underscores from fixtures.
• Rename some remote sample files to make more sense.
• Moves all cli commands/options from cli.py to cli_commands to make it clean and easier to add new commands
• Removes redundant tests
• Updates to new GitHub organisation name in the repo
  • Fixes related links

0.7.0 (2021-09-16)

Major and Feature Improvements

• Drops support for python 3.6
• Update minimum requirement to python 3.7
• Adds support for python 3.9
• Adds models base to the repository. Currently, PyTorch models are supported. New custom models can be added. The tiatoolbox also supports using custom weights to pre-existing built-in models.
  • Adds classification package and CNNPatchPredictor which takes predefined model architecture and pre-trained weights as input. The pre-trained weights for classification using kather100k data set is automatically downloaded if no weights are provided as input.
• Adds mask-based patch extraction functionality to extract patches based on the regions that are highlighted in the input_mask. If 'auto' option is selected, a tissue mask is automatically generated for the input_image using tiatoolbox TissueMasker functionality.
• Adds visualisation module to overlay the results of an algorithm.

Changes to API

• Command line interface for stain normalisation can be called using the keyword stain-norm instead of stainnorm
• Replaces FixedWindowPatchExtractor with SlidingWindowPatchExtractor .
• get_patchextractor takes the slidingwindow as an argument.
• Depreciates VariableWindowPatchExtractor

Bug Fixes and Other Changes

• Significantly improved python notebook documentation for clarity, consistency and ease of use for non-experts.
• Adds detailed installation instructions for Windows, Linux and Mac

Development related changes

• Moves flake8 above pytest in the travis.yml script stage.
• Adds set -e at the start of the script stage in travis.yml to cause it to exit on error and (hopefully) not run later parts of the stage.
• Readthedocs related changes
  • Uses requirements.txt in .readthedocs.yml
  • Uses apt-get installation for openjpeg and openslide
  • Removes conda build on readthedocs build
• Adds extra checks to pre-commit, e.g., import sorting, spellcheck etc. Detailed list can be found on this commit.

0.6.0 (2021-05-11)

Major and Feature Improvements

• Add TissueMasker class to allow tissue masking using Otsu and Morphological processing.
• Add helper/convenience method to WSIReader(s) to produce a mask. Add reader object to allow reading a mask conveniently as if it were a WSI i.e., use same location and resolution to read tissue area and mask area.
• Add PointsPatchExtractor returns patches that can be used by classification models. Takes csv, json or pd.DataFrame and returns patches corresponding to each pixel location.
• Add feature FixedWindowPatchExtractor to run sliding window deep learning algorithms.
• Add example notebooks for patch extraction and tissue masking.
• Update readme with improved instructions to use the toolbox. Make the README file somewhat more comprehensible to beginners, particularly those with not much background or experience.

Changes to API

• tiatoolbox.dataloader replaced by tiatoolbox.wsicore

Bug Fixes and Other Changes

• Minor bug fixes

Development-related changes

• Improve unit test coverage.
• Move test data to tiatoolbox server.

---

0.5.2 (2021-03-12)

Bug Fixes and Other Changes

• Fix URL for downloading test JP2 image.
• Update readme with new logo.

---

0.5.1 (2020-12-31)

Bug Fixes and Other Changes

• Add scikit-image as dependency in setup.py
• Update notebooks to add instructions to install dependencies

---

0.5.0 (2020-12-30)

Major and Feature Improvements

• Adds get_wsireader() to return appropriate WSIReader.
• Adds new functions to allow reading of regions using WSIReader at different resolutions given in units of:
  • microns per-pixel (mpp)
  • objective lens power (power)
  • pixels-per baseline (baseline)
  • resolution level (level)
• Adds functions for reading regions are read_bounds and read_rect.
  • read_bounds takes a tuple (left, top, right, bottom) of coordinates in baseline (level 0) reference frame and returns a region bounded by those.
  • read_rect takes one coordinate in baseline reference frame and an output size in pixels.
• Adds VirtualWSIReader as a subclass of WSIReader which can be used to read visual fields (tiles).
  • VirtualWSIReader accepts ndarray or image path as input.
• Adds MPP fall back to standard TIFF resolution tags with warning.
  • If OpenSlide cannot determine microns per pixel (mpp) from the metadata, checks the TIFF resolution units (TIFF tags: ResolutionUnit, XResolution and YResolution) to calculate MPP. Additionally, add function to estimate missing objective power if MPP is known of derived from TIFF resolution tags.
• Estimates missing objective power from MPP with warning.
• Adds example notebooks for stain normalisation and WSI reader.
• Adds caching to slide info property. This is done by checking if a private self._m_info exists and returning it if so, otherwise self._info is called to create the info for the first time (or to force regenerating) and the result is assigned to self._m_info. This could in future be made much simpler with the functools.cached_property decorator in Python 3.8+.
• Adds pre processing step to stain normalisation where stain matrix encodes colour information from tissue region only.

Changes to API

• read_region refactored to be backwards compatible with openslide arguments.
• slide_info changed to info
• Updates WSIReader which only takes one input
• WSIReader input_path variable changed to input_img
• Adds tile_read_size, tile_objective_value and output_dir to WSIReader.save_tiles()
• Adds tile_read_size as a tuple
• transforms.imresize takes additional arguments output_size and interpolation method 'optimise' which selects cv2.INTER_AREA for scale_factor<1 and cv2.INTER_CUBIC for scale_factor>1

Bug Fixes and Other Changes

• Refactors glymur code to use index slicing instead of deprecated read function.
• Refactors thumbnail code to use read_bounds and be a member of the WSIReader base class.
• Updates README.md to clarify installation instructions.
• Fixes slide_info.py for changes in WSIReader API.
• Fixes save_tiles.py for changes in WSIReader API.
• Updates example_wsiread.ipynb to reflect the changes in WSIReader.
• Adds Google Colab and Kaggle links to allow user to run notebooks directly on colab or kaggle.
• Fixes a bug in taking directory input for stainnorm operation for command line interface.
• Pins numpy<=1.19.3 to avoid compatibility issues with opencv.
• Adds scikit-image or jupyterlab as a dependency.

Development related changes

• Moved test_wsireader_jp2_save_tiles to test_wsireader.py.
• Change recipe in Makefile for coverage to use pytest-cov instead of coverage.
• Runs travis only on PR.
• Adds pre-commit for easy setup of client-side git hooks for black code formatting and flake8 linting.
• Adds flake8-bugbear to pre-commit for catching potential deepsource errors.
• Adds constants for test regions in test_wsireader.py.
• Rearranges usage.rst for better readability.
• Adds pre-commit, flake8, flake8-bugbear, black, pytest-cov and recommonmark as dependency.

---

0.4.0 (2020-10-25)

Major and Feature Improvements

• Adds OpenSlideWSIReader to read Openslide image formats
• Adds support to read Omnyx jp2 images using OmnyxJP2WSIReader.
• New feature added to perform stain normalisation using Ruifork, Reinhard, Vahadane, Macenko methods and using custom stain matrices.
• Adds example notebook to read whole slide images via the toolbox.
• Adds WSIMeta class to save meta data for whole slide images. WSIMeta casts properties to python types. Properties from OpenSlide are returned as string. raw values can always be accessed via slide.raw. Adds data validation e.g., checking that level_count matches up with the length of the level_dimensions and level_downsamples. Adds type hints to WSIMeta.
• Adds exceptions FileNotSupported and MethodNotSupported

Changes to API

• Restructures WSIReader as parent class to allow support to read whole slide images in other formats.
• Adds slide_info as a property of WSIReader
• Updates slide_info type to WSIMeta from dict
• Depreciates support for multiprocessing from within the toolbox. The toolbox is focused on processing single whole slide and standard images. External libraries can be used to run using multiprocessing on multiple files.

Bug Fixes and Other Changes

• Adds scikit-learn, glymur as a dependency
• Adds licence information
• Removes pathos as a dependency
• Updates openslide-python requirement to 1.1.2

---

0.3.0 (2020-07-19)

Major and Feature Improvements

• Adds feature read_region to read a small region from whole slide images
• Adds feature save_tiles to save image tiles from whole slide images
• Adds feature imresize to resize images
• Adds feature transforms.background_composite to avoid creation of black tiles from whole slide images.

Changes to API

• None

Bug Fixes and Other Changes

• Adds pandas as dependency

---

0.2.2 (2020-07-12)

Major and Feature Improvements

• None

Changes to API

• None

Bug Fixes and Other Changes

• Fix command line interface for slide-info feature and travis pypi deployment

---

0.2.1 (2020-07-10)

Major and Feature Improvements

• None

Changes to API

• None

Bug Fixes and Other Changes

• Minor changes to configuration files.

---

0.2.0 (2020-07-10)

Major and Feature Improvements

• Adds feature slide_info to read whole slide images and display meta data information
• Adds multiprocessing decorator TIAMultiProcess to allow running toolbox functions using multiprocessing.

Changes to API

• None

Bug Fixes and Other Changes

• Adds Sphinx Readthedocs support https://readthedocs.org/projects/tia-toolbox/ for stable and develop branches
• Adds code coverage tools to test the pytest coverage of the package
• Adds deepsource integration to highlight and fix bug risks, performance issues etc.
• Adds README to allow users to setup the environment.
• Adds conda and pip requirements instructions

---

0.1.0 (2020-05-28)

• First release on PyPI.