Table 4.

Case-control discriminatory performance of standard breast cancer risk factors combined with parenchymal texture features.

	Breast-anatomy-driven approach		Lattice-based approach 23		p c
	AUC, 95% CI	p a	AUC, 95% CI	p a
Baseline 1 + Texture	0.62, 95% CI [0.56 0.67]	0.011	0.61, 95% CI [0.56 0.64]	0.029	0.051
Baseline 2 + Texture	0.65, 95% CI [0.58 0.69]	0.031	0.60, 95% CI [0.56 0.65]	0.042	0.033
Baseline 3 + Texture	0.66, 95% CI [0.60 0.71]	0.030	0.64, 95% CI [0.60 0.66]	0.037	0.039
Baseline 4 + Texture	0.67, 95% CI [0.60 0.72]	0.029	0.64, 95% CI [0.61 0.66]	0.038	0.027

Cross-validated area under the curve (AUC) and 95% confidence intervals (CIs) for baseline models augmented by breast-anatomy-driven texture features or conventional lattice-based texture descriptors.

p^a: p-value for difference in AUC from the corresponding baseline model; p^c: p-value for difference in AUC between breast-anatomy-driven and lattice-based texture analysis, by DeLong’s tests.