Challenge Tasks
The SegRap2025 Challenge focuses on accurate Gross Tumor Volume (GTV) and Lymph Node Clinical Target Volume (LN CTV) segmentation in Computed Tomography (CT) images, aiming to support clinical manual delineation, enhance research on radiation dose calculation, and improve efficiency of radiotherapy treatment planning.
Task01 (GTV segmentation) emphasizes model generalizability.
Models are supposed to accurately delineate Gross Target Volume of nasopharynx (GTVnx,
also named GTVp)
and Gross Target Volume of lymph node (GTVnd) within paired non-contrast computed
tomography (ncCT) and
contrast computed tomography (ceCT) images.
The testing data are from two cohorts: one internal cohort and one external cohort.
Task02 (LN CTV segmentation) evaluates model's ability to handle missing modality
problem.
Models are supposed to accurately delineate 6 LN CTVs: L_Ib, L_II+III+Va, L_IV+Vb+Vc, R_Ib,
R_II+III+Va, and R_IV+Vb+Vc within paired ncCT and ceCT images or ncCT/ceCT images.
The validation and
testing data are come from one external cohort, comprising cases with: both ncCT and ceCT
scans; only ncCT scans; and only ceCT scans.