Table 7.

Evaluations of segmentation results obtained in axial, coronal, sagittal views before and after post-processing, and evaluations of fusion results before and after post-processing (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	axial	0.85	0.80	0.70	0.87	0.80	0.63	0.84	0.81	0.80
		coronal	0.82	0.80	0.70	0.77	0.78	0.62	0.90	0.83	0.82
		sagittal	0.83	0.80	0.71	0.80	0.79	0.64	0.87	0.83	0.82
	fusing (FCNN + CRF)		0.85	0.83	0.74	0.85	0.84	0.68	0.88	0.82	0.82
	FCNN + CRF + post-process	axial	0.87	0.83	0.76	0.92	0.87	0.77	0.83	0.81	0.77
		coronal	0.87	0.82	0.76	0.86	0.83	0.74	0.89	0.83	0.79
		sagittal	0.87	0.82	0.77	0.88	0.83	0.75	0.86	0.82	0.79
	fusing (FCNN + CRF) + post-process		0.88	0.84	0.77	0.90	0.87	0.76	0.86	0.82	0.80
Learderboard	FCNN + CRF	axial	0.83	0.66	0.57	0.85	0.71	0.50	0.85	0.69	0.71
		coronal	0.80	0.65	0.56	0.75	0.69	0.50	0.90	0.69	0.70
		sagittal	0.80	0.66	0.55	0.78	0.69	0.49	0.86	0.69	0.70
	fusing (FCNN + CRF)		0.84	0.67	0.60	0.84	0.74	0.54	0.87	0.68	0.72
	FCNN + CRF + post-process	axial	0.86	0.73	0.62	0.89	0.76	0.64	0.84	0.78	0.68
		coronal	0.86	0.73	0.62	0.86	0.73	0.60	0.89	0.79	0.67
		sagittal	0.84	0.71	0.60	0.85	0.74	0.59	0.84	0.74	0.67
	fusing (FCNN + CRF) + post-process		0.86	0.73	0.62	0.89	0.77	0.60	0.85	0.77	0.69