MRI is popular in radiation oncology because of its excellent imaging quality of soft tissue and tumor. With the advent of MR-Linac and MR-guided radiation therapy, there is a trend toward a MR-based radiation treatment planning. Contouring is an important task in modern radiation treatment planning and frequently introduces uncertainties in radiation therapy due to observer variabilities. Auto-segmentation has been demonstrated as an effective approach to reduce this uncertainty. The overall objective of this grand challenge is to provide a platform for comparison of various auto-segmentation algorithms when they are used to delineate organs at risk (OARs) or tumors from MR images for head and neck patients for radiation treatment planning. The results will provide an indication of the performances achieved by various auto-segmentation algorithms and can be used to guide the selection of these algorithms for clinic use if desirable.
The data for this challenge contains a total of 55 MRI cases, each from a single examination from a distinct patient, with each case consisting of a T2-weighted MRI images in DICOM format. The MRI scanning protocol was designed for radiation treatment simulation. Thirty-one of these will be provided as training cases, with the parotid glands, submandibular glands, level 2 and level 3 lymph nodes contoured. The images and contours were acquired from MD Anderson Cancer Center.
More details on accessing the various challenge subsets (training, off-site test, and live test) can be found on the Detailed Description tab.
We would like to acknowledge the individuals and institutions that have provided data for this collection: