Figure 3
From: Unsupervised anomaly detection with generative adversarial networks in mammography
Examples of anomaly detection with real mammographic image as an input (first image), one of the most similar nine normal synthetic mammographic images as an output (second image), and difference map between the real image and the synthetic image (last image). (a–d) Show true-positive, false-negative, false-positive, and true-negative case in order.
