AI analytics can be used as imaging biomarkers for predicting invasive upgrade of ductal carcinoma in situ

Yoon, Jiyoung; Yang, Juyeon; Lee, Hye Sun; Kim, Min Jung; Park, Vivian Youngjean; Rho, Miribi; Yoon, Jung Hyun

doi:10.1186/s13244-024-01673-0

Insights into Imaging

Table 3 Multivariable analysis for predictors of invasive upgrade in the total 440 DCIS patients

From: AI analytics can be used as imaging biomarkers for predicting invasive upgrade of ductal carcinoma in situ

Variables		Clinicopathologic variables + mammographic features		Clinicopathologic variables + BI-RADS assessments		Clinicopathologic variables + raw numerical AI-CAD scores		Clinicopathologic features + dichotomized AI-CAD score		Clinicopathologic features + graded AI-CAD score
Variables		OR (95% CI)	p value	OR (95% CI)	p value	OR (95% CI)	p value	OR (95% CI)	p value	OR (95% CI)	p value
Bilateral breast cancer	No	ref		ref		ref		ref		ref
Bilateral breast cancer	Yes	0.527 (0.240–1.157)	0.110	0.558 (0.252–1.235)	0.150	0.533 (0.243–1.167)	0.115	0.508 (0.233–1.107)	0.0883	0.530 (0.242–1.162)	0.113
Related symptoms	No	ref		ref		ref		ref		ref
Related symptoms	Yes	1.290 (0.731–2.278)	0.380	1.222 (0.695–2.148)	0.486	1.348 (0.780–2.331)	0.285	1.331 (0.770–2.300)	0.3054	1.356 (0.782–2.351)	0.279
Biopsy method	Core needle biopsy	ref		ref		ref		ref		ref
Biopsy method	Vacuum-assisted biopsy	0.380 (0.203–0.712)	0.003	0.466 (0.249–0.874)	0.017	0.390 (0.212–0.716)	0.002	0.384 (0.209–0.705)	0.002	0.380 (0.206–0.698)	0.002
Tumor size		1.004 (0.988–1.021)	0.610	0.997 (0.980–1.014)	0.720	1.002 (0.986–1.018)	0.827	1.004 (0.987–1.020)	0.667	1.002 (0.986–1.018)	0.819
Nuclear grade	Low	ref		ref		ref		ref		ref
	Intermediate	1.973 (1.046–3.722)	0.036	1.830 (0.964–3.475)	0.065	1.779 (0.938–3.373)	0.078	1.839 (0.971–3.483)	0.062	1.827 (0.963–3.465)	0.065
	High	2.624 (1.130–6.096)	0.025	2.124 (0.893–5.048)	0.088	2.432 (1.045–5.663)	0.039	2.539 (1.094–5.893)	0.030	2.455 (1.055–5.714)	0.037
Comedonecrosis	No	ref		ref		ref		ref		ref
Comedonecrosis	Yes	0.920 (0.510–1.660)	0.78	0.867 (0.473–1.587)	0.643	0.950 (0.527–1.712)	0.863	0.969 (0.538–1.745)	0.915	0.945 (0.524–1.705)	0.850
Imaging features on mammography	Mass/asymmetry/distortion	ref
	Occult	0.766 (0.330–1.780)	0.536
	Calcifications only	1.553 (0.732–3.294)	0.251
	Combined	2.225 (1.068–4.633)	0.033
Final assessment for mammography	BI-RADS 1–2			ref
	BI-RADS 3			1.419 (0.348–5.786)	0.626
	BI-RADS 4a			1.137 (0.472–2.734)	0.775
	BI-RADS 4b			1.693 (0.719–3.985)	0.228
	BI-RADS 4c			2.473 (1.135–5.390)	0.023
	BI-RADS 5			5.190 (2.175–12.384)	< 0.001
Raw numerical AI-CAD score						1.009 (1.003–1.016)	0.007
Dichotomized AI-CAD score	AI-CAD score < 50%							ref
Dichotomized AI-CAD score	AI-CAD score ≥ 50%							1.960 (1.130–3.399)	0.017
Graded AI-CAD score	AI-CAD score < 25%									ref
	AI-CAD score 25–50%									1.234 (0.489–3.112)	0.657
	AI-CAD score 50–75%									1.260 (0.487–3.264)	0.634
	AI-CAD score ≥ 75%									2.306 (1.233–4.313)	0.009
AUC (95% CI)		0.708 (0.655–0.761)		0.724 (0.671–0.777)		0.702 (0.648–0.755)		0.699 (0.645–0.753)		0.703 (0.649–0.757)

DCIS Ductal carcinoma in situ, OR Odds ratio, 95% CI 95% confidence interval, ref Reference, BI-RADS Breast Imaging Reporting And Data System, AI-CAD AI-based computer-aided detection/diagnosis, AUC Area under curve

Back to article page