Full data for the workshop

This release contains the zip archives with detected cells for the workshop. The data is derived from the PANDa challange (https://www.kaggle.com/competitions/prostate-cancer-grade-assessment/data) by running the CellViT-256-x40 model from CellViT (https://github.com/TIO-IKIM/CellViT) cell detection framework on a subset of all data.
The dataset contains in total 10615 files of the PANDa dataset, which excludes one file from the orginal set which did not contain any information. Incuded are GeoJSON files which contains the detection results as well as the corresponding CSV files which details the data in the split. The splits are roughly 80% train, 10% dev and 10% test.

Test data for the workshop

This contains a subset of the PANDa data. The archive contains GeoJSON files with the detected location of cells in the original whole slide images.

Data from the PANDa dataset for the workshop

This release contains the zip archives with detected cells for the workshop. The data is derived from the PANDa challange (https://www.kaggle.com/competitions/prostate-cancer-grade-assessment/data) by running the CellViT-256-x40 model from CellViT (https://github.com/TIO-IKIM/CellViT) cell detection framework on a subset of all data.
The subset contains in total 1998 files sampled with a balanced bias (i.e. trying to include a uniforma sample of the classes) from the total 10616 files of the PANDa dataset. Incuded are GeoJSON files which contains the detection results as well as the corresponding CSV files which details the included data. The splits are roughly 80% train, 10% dev and 10% test.