Lung Nodule Analysis 2022 - ISMI¶

This is an educational challenge for the 2022 edition of the ISMI course at Radboud University.

📢Only participants of the ISMI course at Radboud University will be given access to participate and submit solutions to this challenge.

This challenge will evaluate lung nodule classifiers on chest CT images. Specifically, the methods will be assessed on volumes of interest (VOI) of size 128 x 128 x 64 voxels (in x, y, and z directions) for malignancy-risk estimation and nodule type classification.

Training dataset¶

We provide VOIs around lung nodules from the publicly available LIDC-IDRI dataset. Following the LUNA16 criteria, we select 1186 nodules annotated by at least 3 out of 4 radiologists. We include 10 nodules for the phase 1 test set, which you can use to test your algorithm submission pipeline. The remaining 1176 nodules are provided as the development dataset. Participants can use this dataset to build their nodule classifiers. 

Testing dataset¶

We provide VOIs around lung nodules from the Danish Lung Cancer Screening Trial for external validation of the submitted algorithms. This dataset contains 177 nodules in total, and this will be available as a hidden test set in phase 2. The algorithm containers will be run on this dataset by the infrastructure available on grand-challenge.org.