Table 4.

Performance of different feature groups.

Feature type	Feature Dimension	Measurements	Dice			PPV			Sensitivity			Specificity
		ROI	Complete	Core	Enhancing	Complete	Core	Enhancing	Complete	Core	Enhancing	Complete	Core	Enhancing
Gradient2D	12	HGG&LGG	0.89	0.54	0.41	0.87	0.47	0.49	0.90	0.63	0.35	0.50	0.56	0.92
		HGG	0.89	0.55	0.40	0.88	0.48	0.49	0.90	0.64	0.34	0.51	0.57	0.91
		LGG	0.85	0.54	–	0.84	0.48	–	0.87	0.62	–	0.47	0.62	–
Gradient3D	34	HGG&LGG	0.90	0.61	0.60	0.92	0.60	0.63	0.89	0.62	0.57	0.71	0.75	0.92
		HGG	0.91	0.62	0.60	0.93	0.61	0.63	0.90	0.63	0.57	0.72	0.76	0.91
		LGG	0.85	0.51	–	0.88	0.50	–	0.83	0.55	–	0.62	0.55	–
Context-sensitive	4	HGG&LGG	0.86	0.51	0.14	0.87	0.44	0.23	0.85	0.60	0.11	0.51	0.48	0.91
		HGG	0.87	0.51	0.16	0.87	0.44	0.24	0.86	0.60	0.12	0.51	0.47	0.90
		LGG	0.79	0.45	–	0.83	0.43	–	0.76	0.47	–	0.52	0.47	–
CCS	12	HGG&LGG	0.86	0.54	0.13	0.89	0.47	0.25	0.84	0.62	0.09	0.62	0.57	0.94
		HGG	0.87	0.53	0.16	0.90	0.48	0.27	0.85	0.61	0.11	0.62	0.56	0.92
		LGG	0.80	0.48	–	0.86	0.47	–	0.75	0.48	–	0.64	0.48	–