Table 2.

Testing results for each scenario in different methods. Numbers in bold font indicate the best performance in each group. 2-D Iter means the iterative method using 2-D convolutions.

		DSC	HD95 (mm)	HD80 (mm)	MSD (mm)	VS
No Prior	Proposed	0.82 ± 0.01	15.16 ± 2.05	5.63 ± 0.94	4.77 ± 0.58	0.93 ± 0.01
	2-D Iter	0.69 ± 0.06	35.36 ± 11.44	22.77 ± 9.66	15.67 ± 6.18	0.84 ± 0.08
	2-D	0.31 ± 0.03	69.68 ± 4.15	47.67 ± 4.64	26.79 ± 2.95	0.65 ± 0.09
	3-D	0.64 ± 0.03	43.03±4.63	27.17 ± 3.58	18.05 ± 3.82	0.76 ± 0.05
Prior in Inference	Proposed	0.88 ± 0.01	3.50 ± 0.51	1.62 ± 0.07	1.79 ± 0.14	0.98 ± 0.01
	2-D Iter	0.70 ± 0.07	33.48 ± 12.70	22.39 ± 10.09	15.49 ± 6.69	0.84 ± 0.09
	2-D	0.37 ± 0.05	72.20 ± 6.54	45.89 ± 9.48	28.01 ± 6.23	0.59 ± 0.06
	3-D	0.70 ± 0.05	37.78 ± 2.51	22.52 ± 4.07	17.46 ± 5.93	0.78 ± 0.05
Prior in Training and Inference	Proposed	0.87 ± 0.01	5.17 ± 1.81	1.98 ± 0.31	2.34 ± 0.31	0.97 ± 0.01
	2-D Iter	0.75 ± 0.02	24.27 ± 4.62	14.56 ± 3.54	11.23 ± 3.48	0.92 ± 0.02
	2-D	0.46 ± 0.08	61.01 ± 4.77	33.07 ± 2.55	21.44 ± 1.93	0.64 ± 0.11
	3-D	0.73 ± 0.02	36.12 ± 2.36	18.75 ± 0.31	12.67 ± 0.43	0.79 ± 0.03