Evaluation of functional magnetic resonance APT and DKI imaging for breast cancer

Table 5 ROC analysis of the diagnostic performance of parameters and methods alone or in combination in distinguishing between benign and malignant breast lesions

Multi parameters	AUC (95%)	P value	Youden index	Sensitivity (%)	Specificity (%)
Parameters
Kapp	0.871 (0.818–0.924)	<0.0001	0.637	98.2	65.5
Dapp	0.872 (0.816–0.928)	<0.0001	0.660	85.5	80.5
MTRasym	0.643 (0.550–0.735)	0.0027	0.342	54.5	79.6
ADC	0.928 (0.888–0.968)	<0.0001	0.703	81.8	88.5
Combined diagnosis
DKI + APT	0.893 (0.843–0.942)	<0.0001	0.676	80.0	87.6
DKI + ADC	0.936 (0.898–0.973)	<0.0001	0.765	81.8	94.7
APT + ADC	0.925 (0.881–0.969)	<0.0001	0.733	92.7	80.5
DKI + APT + ADC	0.933 (0.893–0.974)	<0.0001	0.765	81.8	94.7

AUC: Kapp > MTRasym (Z = −4.191, P <.0001), Dapp > MTRasym (3.5 ppm) (Z = −3.943, P <.0001), ADC > MTRasym (Z = −5.677, P <.0001), ADC > Kapp (Z = −2.182, P = .029), ADC > Dapp (Z = −2.358, P = .018); DKI + ADC > DKI + APT (Z = −2.835, P = .005), DKI + APT > DKI + APT + ADC (Z = −2.947, P = .003). No considerable difference in AUCs of other parameters and methods was noted. ADC, apparent diffusion coefficient; APT, amide proton transfer; AUC, area under the curve; Dapp, apparent diffusivity; DKI, diffusion kurtosis imaging; Kapp, apparent kurtosis coefficient; MTRasym (3.5 ppm), asymmetric magnetization transfer ratio at 3.5 ppm

ISSN: 1475-2867