Here we provide the data sets supporting the experiments in our publication Normalization of HE-Stained Histological Images using Cycle Consistent Generative Adversarial Networks, which were collected at the Institute of Pathology, Medical Faculty Mannheim, Heidelberg University.
The HE-Staining Variation (HEV) data set offers serial sections of a follicular thyroid carcinoma, stained with different HE-staining protocols (including name of [stainVariant]):
- stained with HE with the standard protocol of the Institute of Pathology, Mannheim (HE)
- stained too long with HE (longHE)
- stained too short with HE (shortHE)
- stained only with Hematoxylin (onlyH)
- stained only with Eosin (onlyE)
- stained too long with Hematoxylin (longH)
- stained too long with Eosin (longE)
- stained too short with Hematoxylin (shortH)
- stained too short with Eosin (shortE)
We provided the original whole-slide-images (WSI) in the folder HEV_wsi.zip for each stain-variant.

In addition, for the stain-variants 1-5 we provide patches (n ~40,000 for each set) of size 256x256 pixels and split them into 60% train (train_[stainVariant].zip) and 40% test (test_[stainVariant].zip) sets .

Patches from our TumorLymphnode data set for image classification are provided inside tumorLymphnode_patches.zip. It contains ~3,600 patches of size 165x165 pixels for each class normal lymph nodes (normal) and carcinoma infiltration (tumor).
The code for our models is available at Gitlab.