Skip to main content
Fig. 4 | European Radiology Experimental

Fig. 4

From: Deep learning for predicting future lesion emergence in high-risk breast MRI screening: a feasibility study

Fig. 4

The association of anomaly scores in negative scans (BI-RADS 1) and the emergence of lesions (BI-RADS 2 and above) in future scans. a ROC curves for 10 randomly drawn subsamples of the test set with equal numbers of subjects from the positive and negative cohorts are shown in different colors. Each color corresponds to one of the subsamples. b Illustrative examples of false negatives (low anomaly, future lesion), true positives (high anomaly, future lesion), false positives (high anomaly, no future lesion), and true negatives (low anomaly, no future lesion) in the test set. BI-RADS Breast Imaging and Reporting Data System

Back to article page