H&E staining of human tissue sections is a routine and most common protocol used by pathologists to enhance the contrast of tissue sections for tumor assessment (grading, staging, etc.) at multiple microscopic resolutions. Hence, we will provide the annotated dataset of H&E stained digitized tissue images of several patients acquired at multiple hospitals using one of the most common 40x scanner magnification. The annotations will be done with the help of expert pathologists.
The challenge data is released under the creative commons license (CC BY-NC-SA 4.0).
The training data contains images from 4 different organs (Lung, Prostate, Kidney, and Breast) and it includes images in .svs and .tif format along with 31,000 nuclear boundary annotations in .xml files. The details can be found in the related publication:
- R. Verma, et al. "MoNuSAC2020: A Multi-organ Nuclei Segmentation and Classification Challenge." IEEE Transactions on Medical Imaging (2021).
- The dataset (images and annotations) can be downloaded from the link - MoNuSAC2020 Training Data
- Supplementary document containing organ information is available at the link- Training data organ information
- Code for reading the .xml annotation files can be downloaded by clicking here.
In addition to testing images we also provided binary masks of ambiguous regions that were not included in the challenge metric computation while ranking the MoNuSAC submissions. The ambiguous regions are those that have very faint nuclei with fuzzy boundaries, nuclei for which class assignments were difficult (high chances of incorrect manual labeling) and other nuclei not included in this challenge (endothelial cells, fibroblasts, etc.). You can use the provided ambiguous region masks to remove any nuclei predicted by your algorithm within those regions.
- Testing dataset (images and annotations) can be downloaded using the link- MoNuSAC2020 Testing Data
- Color coded ground truth masks and predictions of top five teams are available below
It should be noted that the aforementioned color-coded images are only for illustrative purposes and should not be used for quantitative assessment of the particpants’ algorithms.