Other Open Access
Nicholas Heller; Nikolaos Papanikolopoulos; Christopher Weight
This is the challenge design document for the "2021 Kidney and Kidney Tumor Segmentation Challenge", accepted for MICCAI 2021.
There is currently a great interest in quantitatively studying the morphology of kidney tumors in order to better characterize surgical complexity and inform treatment planning. Semantic segmentation is a powerful tool for this, but it requires expert reading and considerable manual effort. The KiTS19 challenge introduced the first large-scale public dataset of kidney and kidney tumor semantic segmentations, representing a considerable step towards reliable automatic segmentation of these structures. Unfortunately, it was limited in both the scope of the dataset and the structures that were annotated. The goal of the KiTS21 challenge is to address these limitations by incorporating data from disparate geographical locations and acquisition times, and by providing segmentation labels for more extensive anatomical structures such as the ureters and renal vessels. This will enhance both the clinical utility of the resulting methods, as well as the technical challenge for participants.