Skip to main content
. 2017 Mar 30;12(3):e0174681. doi: 10.1371/journal.pone.0174681

Table 4. Univariate and multivariate logistic regression analysis of DWI characteristics for predicting malignancy.

Lesion characteristics (for mass type) No. of benign lesions (%) No. of malignant lesions (%) Univariate analysis Multivariate analysis
Odds Ratio (95% CI) p-value Odds Ratio (95% CI) p-value
Qualitative DWI analysis 25 (20.7) 96 (79.3)
Shape
oval 14 (53.9) 12 (46.2) 1 1
round 2 (33.3) 4 (66.7) 2.33 (0.36–15.05) 0.373 1.33 (0.05–37.46) 0.869
irregular 9 (10.1) 80 (89.9) 10.37 (3.69–29.16) <0.0001 13.56 (0.64–289.51) 0.095
Margin
circumscribed 13 (54.2) 11 (45.8) 1 1
irregular 12 (16.7) 60 (83.3) 5.91 (2.14–16.29) 0.0006 0.32 (0.01–10.39) 0.523
spiculated 0 (0) 25 (100.0) 0.942 0.966
Internal pattern
homogenous 19 (82.6) 4 (17.4) 1 1
heterogenous 6 (9.38) 58 (90.6) 45.92 (11.70–180.18) <0.0001 21.34 (2.44–186.81) 0.006
rim sign 0 (0) 34 (100.0) 0.926 0.947
Quantitative DWI analysis 22 (18.6) 96 (81.4)
ADC<1.0×10-3mm2/s 5 (5.9) 79 (94.1) 15.80 (5.12–48.74) <0.0001 19.07 (2.79–130.24) 0.003
ADC≥1.0×10-3mm2/s 17 (50.0) 17 (50.0) 1

Note_ Lesion characteristics for NME type lesions were not included in the multivariate analysis because variables were not significant in the univariate analysis.

†Determined with the χ2 test.

‡Determined with logistic regression analysis.