Table 2.

The evaluation results of FCNN + CRF with 5 of 6 post-processing steps (FCNN + CRF + post- x indicates FCNN + CRF with all other post-processing steps except the step x, the sizes of patches used to train FCNNs were 33^*33^*3 and 65^*65^*3 respectively, n = 5, and the number of patches used to train FCNNs was 5000^*5^*20.).

Dataset	Methods	Dice			PPV			Sensitivity
		Complete	Core	Enhancing	Complete	Core	Enhancing	Complete	Core	Enhancing
Challenge	FCNN + CRF + post-process	0.87	0.83	0.76	0.92	0.87	0.77	0.83	0.81	0.77
	FCNN + CRF + post-1	0.87	0.83	0.76	0.92	0.87	0.77	0.83	0.81	0.77
	FCNN + CRF + post-2	0.86	0.83	0.76	0.90	0.87	0.77	0.84	0.81	0.77
	FCNN + CRF + post-3	0.85	0.80	0.72	0.88	0.80	0.69	0.83	0.81	0.77
	FCNN + CRF + post-4	0.86	0.82	0.76	0.92	0.88	0.77	0.82	0.78	0.77
	FCNN + CRF + post-5	0.87	0.83	0.74	0.92	0.87	0.70	0.83	0.81	0.79
	FCNN + CRF + post-6	0.87	0.83	0.76	0.92	0.87	0.77	0.83	0.81	0.77
Leaderboard	FCNN + CRF + post-process	0.86	0.73	0.62	0.89	0.76	0.64	0.84	0.78	0.68
	FCNN + CRF + post-1	0.84	0.71	0.62	0.88	0.75	0.64	0.82	0.77	0.68
	FCNN + CRF + post-2	0.86	0.73	0.62	0.88	0.76	0.64	0.86	0.77	0.68
	FCNN + CRF + post-3	0.85	0.70	0.59	0.87	0.73	0.57	0.84	0.75	0.68
	FCNN + CRF + post-4	0.85	0.71	0.62	0.89	0.77	0.64	0.82	0.74	0.68
	FCNN + CRF + post-5	0.86	0.72	0.59	0.89	0.76	0.57	0.84	0.76	0.71
	FCNN + CRF + post-6	0.86	0.70	0.62	0.89	0.76	0.64	0.84	0.71	0.68