Skip to main content
Donate to the Python Software Foundation or Purchase a PyCharm License to Benefit the PSF! Donate Now

Python package for dealing with whole slide images (.svs) for machine learning, including intuitive, painless patch sampling using OpenSlide, automatic labeling from ImageScope XML annotation files, and functions for saving these patches and their meta data into lightning memory-mapped databases (LMDB) for quick reads.

Project description

Current version

Notice: it is strongly recommended to use py-wsi version >= 1.0.

The current update to py_wsi has added three major improvements which are essential for dealing with very large datasets of .svs images:

  • better memory management
  • error handling
  • functionality to allow for sampling test patches before sampling from all images

See this blog post py_wsi for computer analysis on whole slide .svs images using OpenSlide for help on understanding the relationship between patch and tile sampling. The test patch sampling functionality in this version will also help users to know exactly what they are sampling.

For any early users who have downloaded previous versions of py_wsi (< 1.0) I would strongly suggest downloading the update. Please feel free to submit any issues to the GitHub repository and I will provide help as I am able to.

While suggestions for extra/additional functionality will not be immediately considered, pull requests are welcome.

Introduction to py_wsi

py-wsi provides a series of Python classes and functions which deal with databases of whole slide images (WSI), or Aperio .svs files for machine learning, using Python OpenSlide. py-wsi provides functions to perform patch sampling from .svs files, generation of metadata, and several store options: saving to a lightning memory-mapped database (LMDB), HDF5 files, or disk.

These Python functions deal with whole slide images (WSI), or Aperio .svs files for deep learning, using OpenSlide. py-wsi provides functions to perform patch sampling from .svs files, generation of metadata, and several store options: saving to a lightning memory-mapped database (LMDB), HDF5 files, or disk.

Lim et al. in “An analysis of image storage systems for scalable training of deep neural networks” perform a thorough evaluation of the best image storage systems, taking into consideration memory usage and access speed. LMDB, a B+tree based key-value storage, is not the most memory efficient, but provides optimal read time.

py-wsi uses OpenSlide Python. According to the Python OpenSlide website, “OpenSlide is a C library that provides a simple interface for reading whole-slide images, also known as virtual slides, which are high-resolution images used in digital pathology. These images can occupy tens of gigabytes when uncompressed, and so cannot be easily read using standard tools or libraries, which are designed for images that can be comfortably uncompressed into RAM. Whole-slide images are typically multi-resolution; OpenSlide allows reading a small amount of image data at the resolution closest to a desired zoom level.”

Note: HDF5 functionality will not be available until version 1.2

Check Jupyter Notebook on GitHub to view example usage:Example usage of py-wsi

Setup

This library is dependent on the following, but may be compatible with previous versions.

python 3.6.1 numpy 1.12.1 openslide-python 1.1.1

  1. Check dependencies listed in setup.py; notably, openslide-python which requires openslide, and lmdb. The python geometry package Shapely is used for inferring labels from XML annotations.
brew install openslide
  1. Install py_wsi using pip.
pip install py_wsi
  1. Check out Jupyter Notebook “Using py-wsi” to see what py-wsi can do and get started!

Feel free to contact me with any issues and feedback.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Filename, size & hash SHA256 hash help File type Python version Upload date
py_wsi-1.1.tar.gz (12.2 kB) Copy SHA256 hash SHA256 Source None

Supported by

Elastic Elastic Search Pingdom Pingdom Monitoring Google Google BigQuery Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN SignalFx SignalFx Supporter DigiCert DigiCert EV certificate StatusPage StatusPage Status page