Synthetic histology images of colorectal cancer, generated by conditional generative adversarial networks

Jeremias Krause, Jakob Nikolas KatherZenodo 2020 · DOI · OpenAlex

Abstract

These are generated (synthetic) histology images of colorectal cancer. These images were generated by conditional GANs and are in two classes: MSIH (microsatellite instable high) and nonMSIH. There are two sets: one set with 10K images per class and another one with 75K images per class. All images are RGB, 512x512 px at a resolution of 0.5 micrometers per pixel. For more information, please stay tuned for our upcoming manuscript on www.kather.ai.
BibTeX
@inproceedings{krause2020synthetic,
  author = {Jeremias Krause and Jakob Nikolas Kather},
  title = {{Synthetic histology images of colorectal cancer, generated by conditional generative adversarial networks}},
  booktitle = {Zenodo},
  year = {2020},
  doi = {10.5281/zenodo.3955045},
}

Loading related content...

Figure