Table 5.

Area Under Curve (AUC) and Standard Error (SE) of AUC for the ROC Curves for Differentiation of Breast Lesions Based on the Values of ADC₆₀₀, D_d, f_p and the Combination (D_d, f_p), Optimal Sensitivity and Specificity for Each Parameter and the Corresponding Parameter Thresholds

Parameter	AUC (SE)	Sensitivity	Specificity	Threshold
Malignant lesions versus benign lesions
ADC600 (mm2/s)	0.72 (0.08)	0.65	0.71	1.54 ×10−3
Dd(mm2/s)	0.75 (0.08)	0.85	0.64	1.52 × 10−3
fp (%)	0.79 (0.07)	0.73	0.86	4.9
(Dd, fp) (probability, unitless)	0.84 (0.06)	0.85	0.86	0.58
Malignant lesions versus FGTM
ADC600 (mm2/s)	0.86 (0.05)	0.77	0.81	1.57 × 10−3
Dd (mm2/s)	0.90 (0.04)*	0.81	0.85	1.48 × 10−3
fp (%)	0.93 (0.04)	0.88	1.00	3.9
(Dd, fp) (probability, unitless)	0.97 (0.03)†	0.92	0.96	0.40
Benign lesions versus FGTB
ADC600 (mm2/s)	0.69 (0.10)	0.86	0.50	1.91 × 10−3
Dd (mm2/s)	0.69 (0.10)	0.93	0.50	1.88 × 10−3
fp (%)	0.63 (0.11)	0.64	0.79	2.1
(Dd, fp) (probability, unitless)	0.74 (0.10)	0.71	0.64	0.49

FGT_M and FGT_B are FGT ROIs in patients with malignant and benign lesions, respectively. The probability of belonging to the lesion of interest class based on (D_d, f_p) ranges between P = 1 (definitely tissue of interest) and P = 0 (definitely other).

^*AUC(D_d) versus AUC(ADC₆₀₀): P = 0.008.

^†AUC(D_d, f_p) versus AUC(ADC₆₀₀): P = 0.045.