Table 3.

Task 2 (mid-radiotherapy segmentation) results for participating teams. Results are shown in descending order by mean aggregated DSC (DSCagg). DSCagg scores for primary gross tumor volume (GTVp) and metastatic lymph nodes (GTVn) are also shown. Highest scores for each category are bolded. The performance of the mean, baseline nnU-Net, null algorithm (simple structure propagation from registered images), and interobserver variability (derived in section 2.3) are also shown at the bottom of the table. Values are rounded to the nearest 3rd decimal place.

Position	Team name	DSCagg mean	DSCagg GTVp	DSCagg GTVn
1	UW LAIR	0.733	0.607	0.859
2	mic-dkfz	0.727	0.579	0.875
3	HiLab	0.725	0.579	0.871
4	andrei.iantsen	0.718	0.592	0.845
5	Stockholm_Trio	0.710	0.554	0.866
6	lWM	0.707	0.579	0.836
7	RUG_UMCG	0.701	0.543	0.859
8	TUMOR	0.700	0.549	0.852
9	DCPT-Stine’s group	0.700	0.534	0.867
10	alpinists	0.698	0.539	0.858
11	NeuralRad	0.685	0.526	0.843
12	dlabella29	0.655	0.499	0.811
13	CEMRG	0.654	0.534	0.773
14	SJTU&NINTH PEOPLE’S HOSPITAL	0.638	0.446	0.831
15	TNL_skd	0.562	0.500	0.625
NA	Mean	0.688	0.544	0.831
NA	nnU-Net baseline	0.633	0.422	0.844
NA	null algorithm	0.601	0.460	0.743
NA	Interobserver variability	0.714	0.600	0.828