Table 2. Table 2. Usefulness of extracting features from the co-occurrence matrix subwindows.

AUC	HR	HRsub	GR	GRsub	CU	CUsub	LD	LDsub	SH	SHsub
PAP	89.4	92.0	84.1	86.0	77.8	79.0	83.7	82.6	82.5	86.6
VIR	95.9	96.7	88.8	93.4	70.9	70.0	61.3	70.7	84.6	89.9
HI	87.7	88.5	87.3	88.9	67.5	65.6	–	–	83.2	87.6
BC	92.7	93.6	84.9	91.9	83.1	85.1	61.6	85.7	88.8	91.8
PR	90.6	91.0	85.7	91.7	74.8	75.3	80.4	83.5	84.8	84.7
CHO	99.4	99.4	98.6	97.9	98.6	97.8	99.6	99.5	99.5	99.6
Av	92.6	93.5	88.2	91.6	78.7	78.8	–	–	87.2	90.0

Due to huge memory requirements, LD is not performed on HI dataset.