Deep Learning Utilities for Pathology

Dlup offers a set of utilities to ease the process of running Deep Learning algorithms on Whole Slide Images.

Features

• Read whole-slide images at any arbitrary resolution by seamlessly interpolating between the pyramidal levels
• Supports multiple backends, including OpenSlide and VIPS, with the possibility to add custom backends
• Dataset classes to handle whole-slide images in a tile-by-tile manner compatible with pytorch
• Annotation classes which can load GeoJSON, V7 Darwin, HALO and ASAP formats and read parts of it (e.g. a tile)
• Transforms to handle annotations per tile, resulting, together with the dataset classes a dataset consisting of tiles of whole-slide images with corresponding masks as targets, readily useable with a pytorch dataloader
• Command-line utilities to report on the metadata of WSIs, and convert masks to polygons

Check the full documentation for more details on how to use dlup.

Quickstart

The package can be installed using python -m pip install dlup.

Used by

• ahcore: a pytorch lightning based-library for computational pathology

Citing DLUP

If you use DLUP in your research, please use the following BiBTeX entry:

@software{dlup,
  author = {Teuwen, J., Romor, L., Pai, A., Schirris, Y., Marcus, E.},
  month = {8},
  title = {{DLUP: Deep Learning Utilities for Pathology}},
  url = {https://github.com/NKI-AI/dlup},
  version = {0.7.0},
  year = {2024}
}

or the following plain bibliography:

Teuwen, J., Romor, L., Pai, A., Schirris, Y., Marcus E. (2024). DLUP: Deep Learning Utilities for Pathology (Version 0.7.0) [Computer software]. https://github.com/NKI-AI/dlup

Contributors

In alphabetic order:

Ajey Pai Karkala	Eric Marcus	Jonas Teuwen	Leonardo Romor	Rolf Harkes	Yoni Schirris