Getting the data

The dataset is available from a google drive link:

  1. Google drive: [link]
  2. Zenodo (PNG version, due to size restrictions): [link]

Dataset description

After downloading the dataset, you will have 405 PNG files (cropped regions of interest from 455 individual tumor cases), and a database in two formats.The database is available as sqlite (SlideRunner format) and as json (MS COCO format). For an example of how to interpret this format, please see the notebook below.

The assignment of the tumor cases to the files is as follows:

CasesCancer typeSpeciesOther information (scanner, image source, etc.)
001.tiff to 150.tiffbreast cancerhumanUMC Utrecht, scanned with three different scanners (see MIDOG 2021)
151.tiff to 194.tifflung carcinomacanineVetMedUni Vienna, scanned with 3DHistech Pannoramic Scan II
195.tiff to 249.tifflymphomacanineVetMedUni Vienna, scanned with 3DHistech Pannoramic Scan II
250.tiff to 299.tiffmast cell tumorcanineFU Berlin, scanned with Aperio ScanScope CS2
300.tiff to 354.tiffneuroendocrine tumorhumanUMC Utrecht, scanned with Hamamatsu XR
355.tiff to 405.tiffmelanomahumanUMC Utrecht, scanned with Hamamatsu XR (no labels provided for this domain)

In total, the training set contains 9501 mitotic figures and 11051 non-mitotic figure annotations (hard negatives). The hard negatives are provided as they are part of our annotation process and may be helpful within sampling schemes.

Getting started: The notebook

To get you started, there is an explanatory notebook on Google Colab that we really recommend for first steps:

https://colab.research.google.com/drive/1m3zPFWnVbwgPobdsVg1upHCKSl2qQsac?usp=sharing

The notebook comprises the following:

  • Statistical overview of the data set
  • An in-depth view into the dataset

8 Comments

    • Dear Sen,

      sorry for not replying any sooner. Yes, please sign up to the MIDOG 2022 challenge on grand-challenge.org and you are good to go (since you are a verified user of grand-challenge.org, you can sign up without further approval).

      Best,

      Marc

Leave a Reply