Large-scale crowdsourced radiotherapy segmentations across a variety of cancer sites

Clinician generated segmentation of tumors and healthy tissue regions of interest (ROIs) on medical images is crucial for radiotherapy workflows. However, interobserver segmentation variability has long been considered a significant detriment to the implementation of high-quality and consistent radiotherapy dose delivery. This has prompted the increasing development and utilization of computational approaches for automated ROI segmentation, in particular deep learning based methods. However, extant public segmentation datasets typically only provide segmentations generated by a limited number of annotators with varying, and often unspecified, levels of expertise. In this data descriptor, a large number of clinician annotators manually generated segmentations for numerous ROIs on computed tomography images across a variety of cancer sites (breast, sarcoma, head and neck, gynecologic, gastrointestinal) for the Contouring Collaborative for Consensus in Radiation Oncology challenge. Over 200 clinician annotators (verified experts and non-experts) contributed to the generation of these segmentation data using a standardized annotation platform (ProKnow). Subsequently, we converted image and segmentation data into Neuroimaging Informatics Technology Initiative format with standardized nomenclature for ease of use by the research community. In addition, we generated consensus segmentations for experts and non-experts using the simultaneous truth and performance level estimation method. These standardized, structured, and easily accessible data are a valuable resource for systematically studying interobserver variability in ROI segmentation applications with an unprecedented large number of clinician annotators.

Version history details:
1. v1 uploaded 12.10.2022.
2. v2 uploaded 24.01.2023. Some ROI files had names with typos. This was rectified in this version.
3. v3 uploaded 01.02.2023. Modified some file names to increase usability. Added DICOM and NIfTI data for select cases where available (sarcoma - MRI, H&N and GI - PET).
4. v4 uploaded 16.08.2023. Excel file had some empty columns that caused the sheets to go to infinity. Fixed this.