These values were higher than those of BI-RADS when using the threshold > 3 (0.368 [95% CI: 0.284–0.459] in training set and 0.389 [95% CI: 0.231–0.565] in test set), and led to reductions of 12.8% (16/125) and 25% (9/36) unnecessary biopsies, although the specificity improvements did not reach statistical significance via the McNemar test ( p = 0.068 and p = 0.078, respectively).
← all excerpts
Nomogram for reducing unnecessary biopsies of breast lesions based on MRI and clinical features: a multi-center retrospective cohort study.
1
0.0680
0.0680