Table 4.

Comparison evaluation on superpixel classification using SVM-based and ERT-based classifier, respectively, on the 5 features selected using mRMR

Case no.	Grade	SVM			ERT
		Precision (%)	Sensitivity (%)	BER	Precision (%)	Sensitivity (%)	BER
1	II	62.71	97.33	0.02	69.85	97.45	0.02
2	II	58.65	98.14	0.02	90.24	98.65	0.01
3	II	72.55	98.41	0.02	74.21	99.12	0.01
4	II	68.53	94.88	0.03	70.24	96.05	0.02
5	II	76.33	55.64	0.23	78.43	56.32	0.22
6	II	75.83	73.45	0.14	85.63	71.32	0.15
7	III	84.75	98.75	0.01	86.07	99.35	0.01
8	III	88.54	83.32	0.09	90.78	85.64	0.08
9	III	88.92	98.11	0.01	91.44	98.67	0.01
10	IV	95.22	83.25	0.09	97.44	89.03	0.06
11	IV	93.45	88.53	0.07	96.57	91.65	0.05
12	IV	81.55	73.98	0.14	84.33	75.92	0.13
13	IV	80.35	92.68	0.04	82.53	95.73	0.03
14	IV	90.12	92.51	0.04	91.32	95.87	0.03
15	IV	93.42	93.76	0.04	96.78	94.02	0.03
16	IV	87.45	83.06	0.09	90.21	84.15	0.08
17	IV	95.34	87.75	0.06	96.81	91.87	0.04
18	IV	98.33	82.56	0.09	98.43	85.33	0.08
19	IV	96.21	92.51	0.05	98.12	94.03	0.04
Mean	All	83.59	87.82	0.07	87.86	89.48	0.06
STD	All	11.76	11.09	0.06	9.27	11.23	0.06

BER balanced error rate

The classification is performed for tumour including oedema and active tumour core versus normal brain tissue