Diffusion tensor imaging for characterizing tumor microstructure and improving diagnostic performance on breast MRI: a prospective observational study

Table 4 Exploratory modeling for the association of DTI parameters with malignancy in different lesion types

DTI variable	Pathology status				Univariate model			Multivariate model*
	Malignant		Benign		Univariate model			Multivariate model*
	No.	Mean ± SD	No.	Mean ± SD	OR	(95% CI)	p value	OR	(95% CI)	p value
ADC, 10⁻³ mm²/s
Mass	54	1.19 ± 0.27	81	1.59 ± 0.30	0.19	(0.10–0.36)	< 0.001	0.23	(0.13–0.43)	< 0.001
Non-mass	41	1.35 ± 0.36	62	1.49 ± 0.30	0.64	(0.40–1.02)	0.060	0.42	(0.22–0.84)	0.013
					(p = 0.002^†)			(p = 0.20^†)
FA
Mass	54	0.31 ± 0.17	81	0.21 ± 0.12	2.02	(1.36–3.01)	< 0.001	1.62	(0.97–2.71)	0.067
Non-mass	41	0.24 ± 0.11	62	0.26 ± 0.14	0.94	(0.64–1.38)	0.74	0.54	(0.29–1.01)	0.053
					(p = 0.006^†)			(p = 0.009^†)

*The multivariate model included ADC, FA, mass vs. non-mass type, log(lesion size), and interactions between mass/non-mass and ADC and FA; ORs were scaled to show change per 1-SD increase in the corresponding variable; FA was log-transformed prior to inclusion in the logistic regression model to reduce right-skewness
^†Wald test comparing the ORs corresponding to masses and non-masses