A clinical radiomics nomogram preoperatively to predict ductal carcinoma in situ with microinvasion in women with biopsy-confirmed ductal carcinoma in situ: a preliminary study

Table 4 Predictive performances of the six models and DCE-MRI radiomics model in distinguishing DCISMI-more from DICSMI-one

Feature number	Models	Method	AUC(95% CI)	SEN (95% CI%)	SPE (95% CI%)	ACC	PPV	NPV
2	Clinicopathologic	test set	0.76(0.59–0.94)	0.93(0.7–0.99)	0.5(0.27–0.73)	0.72	0.67	0.88
2	Clinicopathologic	training set	0.8(0.69–0.91)	0.87(0.7–0.95)	0.57(0.39–0.73)	0.72	0.68	0.8
2	Conventional MRI	test set	0.77(0.59–0.95)	0.93(0.7–0.99)	0.57(0.33–0.79)	0.76	0.7	0.89
2	Conventional MRI	training set	0.82(0.71–0.93)	0.73(0.56–0.86)	0.82(0.64–0.92)	0.78	0.81	0.74
11	DCE-MRI radiomics	test set	0.74(0.55–0.93)	0.73(0.48–0.89)	0.71(0.45–0.88)	0.72	0.73	0.71
11	DCE-MRI radiomics	training set	0.9(0.83–0.98)	0.87(0.7–0.95)	0.82(0.64–0.92)	0.84	0.84	0.85
13	^aCombine	test set	0.8 (0.63–0.96)	0.73(0.48–0.89)	0.64(0.3–0.84)	0.69	0.69	0.69
13	^aCombine	training set	0.94(0.88–1)	0.93(0.79–0.99)	0.89(0.73–0.96)	0.91	0.9	0.93
4	^bTraditional	test set	0.87(0.73–1)	1(0.8–1)	0.36(0.16–0.61)	0.69	0.63	1
4	^bTraditional	training set	0.91(0.84–0.99)	0.97(0.83–1)	0.82(0.64–0.92)	0.9	0.85	0.96
15	^cMixed	test set	0.93(0.84–1)	0.93(0.7–0.99)	0.5(0.27–0.73)	0.72	0.67	0.88
15	^cMixed	training set	0.98(0.96–1)	0.93(0.79–0.99)	0.93(0.77–0.99)	0.93	0.93	0.93
16	DCE-MRI radiomics predicting DCISMI-more	test set	0.72(0.37–1)	0.67(0.12–0.98)	0.83(0.55–0.97)	0.8	0.5	0.91
16	DCE-MRI radiomics predicting DCISMI-more	training set	1(1–1)	1(0.65–1)	1(0.86–1)	1	1	1

Abbreviations: DCISMI Ductal carcinoma in situ with microinvasive, DCISMI-more DCIS with multifocal of microinvasive carcinoma, DCISMI-one DCIS with one focus of microinvasive carcinoma, AUC Area under the curve, SEN Sensitivity, SPE Specificity, ACC accuracy, PPV Positive predictive value, NPV Negative predictive value, CI Confidence interval, DCE-MRI Dynamic-contrast enhanced MRI
^aCombine model was constructed based on conventional MRI and DCE-MRI radiomics features
^bTraditional model was constructed based on clinicopathologic and conventional MRI features
^cMixed model was constructed based on clinicopathologic, conventional MRI and DCE-MRI radiomics features

ISSN: 1471-2342