Multi-Scanner Canine Cutaneous Squamous Cell Carcinoma Histopathology Dataset

This is a multi-scanner version of the SCC subset of the publicly available CATCH dataset (https://doi.org/10.7937/TCIA.2M93-FX66). The dataset contains 44 samples digitized with five slide scanning systems (220 images)  :

• Aperio ScanScope CS2 (Leica, Germany), resolution: 0.25 µm/pixel
• NanoZoomer S210 (Hamamatsu, Japan), resolution: 0.22 µm/pixel
• NanoZoomer 2.0-HT (Hamamatsu, Japan), resolution: 0.23 µm/pixel
• Pannoramic 1000 (3DHISTECH, Hungary), resolution: 0.25 µm/pixel
• Aperio GT 450 (Leica, Germany), resolution: 0.26 µm/pixel

All Aperio ScanScope CS2 whole slide images were manually annotated for tumor and six skin tissue classes (epidermis, dermis, subcutis, bone, cartilage, and a combined class of inflammation and necrosis). All 1,243 polygon annotations were transferred to the remaining four scanners and are provided as MS COCO JSON and as SQLITE database (SlideRunner format). Due to size restrictions on Zenodo, we provide the files as lower-resolution pyramidal TIFFs (4 µm/pixel).

If you use the dataset for research purposes, please cite our paper:

Frauke Wilm, Marco Fragoso, Christof A. Bertram, Nikolas Stathonikos, Mathias Öttl, Jingna Qiu, Robert Klopfleisch, Andreas Maier, Katharina Breininger, and Marc Aubreville, "Multi-Scanner Canine Cutaneous Squamous Cell Carcinoma Histopathology Dataset", 2023, https://arxiv.org/abs/2301.04423