TABLE 2.

Performance of the selected features for predicting the Ki-67 level and HER-2 status.

Biomarkers	Features	Cohorts	Mean ± SD		AUC	P
Ki-67	lbp-3D-m1_firstorder_InterquartileRange	Training	6.59 ± 1.12	5.85 ± 1.49	0.686	*0.008
		Validation	6.30 ± 1.83	7.02 ± 1.02	0.702	*0.038
	log-sigma-1-0-mm-3D_glcm_InverseVariance	Training	0.41 ± 0.04	0.43 ± 0.03	0.650	*0.035
		Validation	0.41 ± 0.05	0.42 ± 0.02	0.528	0.788
	logarithm_glszm_SmallAreaEmphasis	Training	0.48 ± 0.15	0.56 ± 0.12	0.659	*0.025
		Validation	0.58 ± 0.11	0.54 ± 0.12	0.580	0.420
	wavelet-HHH_ngtdm_Contrast	Training	0.11 ± 0.03	0.12 ± 0.01	0.672	*0.015
		Validation	0.12 ± 0.01	0.11 ± 0.03	0.565	0.617
	wavelet-LHL_firstorder_Skewness	Training	−0.46 ± 0.39	−0.18 ± 0.50	0.699	*0.005
		Validation	−0.62 ± 0.45	−0.24 ± 0.40	0.744	*0.013
HER-2	lbp-3D-k_firstorder_Skewness	Training	0.92 ± 0.42	1.18 ± 0.28	0.706	*0.004
		Validation	1.120 ± 0.45	0.97 ± 0.65	0.660	0.109
	logarithm_gldm_LowGrayLevelEmphasis	Training	0.01 ± 0.02	0.01 ± 0.01	0.664	*0.022
		Validation	0.01 ± 0.02	0.02 ± 0.02	0.667	0.095

SD, standard deviation; AUC, area under the ROC curve; Ki-67, antigen identified by monoclonal antibody; HER-2, human epidermal growth factor receptor 2.

*Statistically significant values of p < 0.05.