Skip to main content
. 2017 May 18;7:2115. doi: 10.1038/s41598-017-02341-8

Table 4.

AUCs of the unconditional logistic regression analyses for testing the BPE measures computed from three different SUBs on distinguishing cancer cases from controls.

SUB1 SUB2 SUB3 SUBs 1, 2, and 3 combined *P-values
Comparison A: Contralateral breasts of both cancer cases and controls
|BPE| 0.612 0.615 0.609 0.612 all p > 0.59
BPE% 0.639 0.654 0.629 0.657 all p > 0.36
|BPE| + BPE% 0.652 0.665 0.663 0.673 all p > 0.41
+P-values all p > 0.29 all p > 0.22 all p > 0.016 all p > 0.12
Comparison B: Benign breast of controls vs contralateral breasts of cancer cases
|BPE| 0.595 0.606 0.595 0.611 all p > 0.20
BPE% 0.614 0.653 0.624 0.654 all p > 0.25
|BPE| + BPE% 0.649 0.669 0.667 0.680 all p > 0.32
+P-values all p > 0.061 all p > 0.12 all p > 0.0047 all p > 0.11

|BPE| = Volume of background parenchymal enhancement.

BPE% = Percentage of background parenchymal enhancement volume (|BPE|) relative to breast volume.

SUB1, SUB2, SUB3 = Subtraction sequence (i.e., post-contrast – pre-contrast) for each of first, second, and third post-contrast sequences, respectively.

*P-values represent the DeLong’s test between any pair of the AUCs with respect to SUB1, SUB2, SUB3, and their combination (i.e., columns 2–5).

+P-values represent the DeLong’s test between any pair of the AUCs with respect to |BPE|, BPE%, and their combination (i.e., rows 3–5 for Comparison A and rows 8–10 for Comparison B).