Optimizing automated characterization of liver fibrosis histological images by investigating color spaces at different resolutions

Table 1 Texture features at full resolution

TA Method	Greylevel	RGB	HSI
COM	Sum of Squares	G_ Sum of Squares	H_ Sum Variance
	Sum Variance	R_ Sum of Squares	H_Correlation
	Sum Entropy	G_ Sum Variance	H_Inverse Difference Moment
RLM	Horizontal greylevel non-uniformity	G_ Horizontal greylevel non-uniformity	I_ Horizontal Run length non-uniformity
	Vertical greylevel non-uniformity	G_45° greylevel non-uniformity	I _Horizontal Fraction
	135°greylevel non-uniformity	G_135° greylevel non-uniformity	I _135° Run length non-uniformity
WT	E ₄	G_E₄	I _E₄
	E ₅	G_E₅	S_E₄
	E ₂	B_E₄	I _E₄

The texture features (parameters with the highest F-Coefficient) that discriminate between the C and F groups on greylevel-, RGB-, and HSI- schemes at full-resolution images, using TA methods:COM, RLM, and WT.
R_: Red, G_: Green, and B_: Blue channels. H_: Hue, S_: Saturation, and I _: Intensity. E_s Energy calculated from the wavelets using various scales (s).

ISSN: 1742-4682