ScribbleDose Annotations: Scribble and Supervoxel Maps for GDP-HMM AAPM Radiotherapy Dose Prediction Data

This dataset accompanies the paper "ScribbleDose: Scribble-Guided Dose Prediction in Radiotherapy". This release provides the processed ScribbleDose annotations for radiotherapy dose prediction. The annotations are generated based on the structure masks from the GDP-HMM AAPM Challenge dataset and include scribble annotations and auxiliary supervoxel maps. To avoid redistributing the original radiotherapy data, this release only contains the processed scribble annotations and supervoxel maps. The original CT images, dose distributions, beam-related information, and full structure masks are not included. Users should first download the original GDP-HMM AAPM Challenge data and then merge the files in this release with the original .npz files using the provided merging script in the ScribbleDose GitHub repository. The released files keep the same filenames as the original GDP-HMM .npz files for data alignment. After merging, the resulting .npz files can be directly used with the ScribbleDose dataloader. GitHub repository: https://github.com/iCherishxixixi/ScribbleDose