Table 4.

Univariate and multivariate logistic regression of the clinical model and combined model for the TN subtype of breast cancer.

Feature	TN	Non-TN	Univariate analysis		Multivariate analysis		With radiomics signature
			P value	Odds ratio	P value	Odds ratio	P value	Odds ratio
Age	54.08 ± 10.48	54.18 ± 9.870	0.954	0.999 (0.965, 1.034)
Size	33.98 ± 17.45	33.35 ± 19.07	0.844	1.002 (0.983, 1.020)
Breast composition
Dense Fatty	40 10	71 29	Ref 0.239	1 0.612 (0.260, 1.351)
Gross feature
Mass only Mass + calcification Calcification only	29 21 0	46 47 7	Ref 0.330 0.986	1 0.709 (0.351, 1.413) NA
Shape
Oval Round Irregular	3 17 30	4 12 77	0.409 0.003 Ref	1.925 (0.362, 9.235) 3.636 (1.567, 8.696) 1	0.016	3.028 (1.233, 7.681)	0.335	1.695 (0.575, 4.998)
Mass margin
Obscured Microlobulated Indistinct Spiculated	10 7 28 5	18 9 50 16	0.986 0.555 Ref 0.301	0.992 (0.393, 2.414) 1.389 (0.452, 4.134) 1 0.558 (0.168, 1.598)
Mass density
Low Equal High	3 23 24	7 61 24	0.861 Ref 0.013	1.137 (0.230, 4.478) 1 2.546 (1.223, 5.372)	0.018	2.542 (1.180, 5.573)	0.370	1.525 (0.598, 3.834)
Architectural distortion	5	25	0.036	0.333 (0.107, 0.869)	0.107	0.403 (0.121, 1.143)	0.419	0.575 (0.138, 2.084)
Calcification morphology
Benign Amorphous Coarse heterogeneous Fine pleomorphic Fine linear branching	1 2 3 11 4	1 2 5 33 14	0.451 0.300 0.468 Ref 0.817	3.000 (0.112, 80.288) 3.000 (0.329, 27.556) 1.800 (0.327, 8.635) 1 0.857 (0.209, 3.009)
Calcification distribution
Diffuse Regional Grouped Linear Segmental	0 2 4 1 14	1 2 13 0 39	0.992 0.328 0.813 0.991 Ref	NA 2.786 (0.310, 25.086) 0.857 (0.214, 2.907) NA 1
Radiomics signature			< 0.001	1781 (190, 23,225)			< 0.001	828 (78, 12,147)

The 95% confidence intervals of the AUCs are shown in parentheses.

TN triple-negative, AUC area under the receiver operating characteristic curve.