Skip to main content
. 2021 Aug 4;11:15857. doi: 10.1038/s41598-021-94750-z

Figure 1.

Figure 1

Synthetic radiographs, ground truth masks and results of reader and computer-based detection. (A) Synthetic input radiograph as shown to the reader and evaluated by the CNNs (B) corresponding ground-truth radiograph with nodules marked green (C) center position of nodules marked by a reader (D) U-Net prediction (E) RetinaNet bounding-box predictions with scores.