PET/CT data for PDL auto-segmentation project

Data for paper corresponding to: Application of simultaneous uncertainty quantification and segmentation for oropharyngeal cancer use-case with Bayesian deep learning. Paper link: https://www.medrxiv.org/content/10.1101/2023.02.20.23286188v1 (will be updated with final manuscript when published).
This dataset is comprised of 67 (anonymized) head and neck cancer cases from MD Anderson Cancer Center. It was used as an external test set for the project. Each case contains a CT image, PET image, and primary gross tumor volume segmentation (binary mask), all in NIfTI format. These data were used as an independent test set for the deep learning algorithms in our study. The images and masks were cropped to be consistent with the 2021 HECKTOR Challenge data formatting. Notably, these cases were used in part for the MDA contribution to the 2022 edition of the HECKTOR Challenge.
In addition, two CSV files are provided for deep learning model outputs generated from the development and external datasets for this project.