Table 6.

All features whose selection frequency is significantly different from the chance in the background/lesion classification (p-value null model test <0.05). Direction and level of the Haar decomposition, and GLCM direction of the extracted features are shown.

Feature Set	Feature	ROI Type	Frequency (%)
HAAR	Mean_LL2	RC	$41.0$
GLCM	SumEntropy_LH1 ($\theta ={90}^{\circ }$)	RC	$40.6$
GLCM	Entropy_LH1 ($\theta ={135}^{\circ }$)	RC	$25.4$
HAAR	Mean_LL2	LE	$23.8$
HAAR	Variance_LH1	RC	$23.6$
GLCM	Entropy_LH1 ($\theta ={90}^{\circ }$)	RC	$22.5$
HAAR	Entropy_LH1	RC	$22.3$
GLCM	SumEntropy_LH1 ($\theta ={45}^{\circ }$)	RC	$21.2$
HAAR	RelativeSmoothness_LH1	RC	$18.4$
HAAR	RelativeSmoothness_LH2	RC	$17.8$
HAAR	Variance_LH2	RC	$17.7$
GLCM	Energy_LH1 ($\theta ={135}^{\circ }$)	RC	$17.5$
HAAR	Mean_LL1	RC	$17.4$
HAAR	Variance_LL2	RC	$17.4$
HAAR	Mean_LL1	LE	$17.2$
GLCM	Entropy_LH1 ($\theta =0^{\circ }$)	RC	$14.6$
STAT	Minimum	RC	$14.5$
GLCM	ClusterProminence_HH1 ($\theta =0^{\circ }$)	RC	$14.3$
COUNT	MSER	RC	$13.5$
HAAR	Variance_LL1	RC	$13.2$
HAAR	RelativeSmoothness_LL1	RC	$12.6$
GLCM	Entropy_LH1 ($\theta ={45}^{\circ }$)	RC	$12.5$