Skip to main content

Table 1 Lesion agreement. Three readers independently annotated lung cancer lesions in the test images. The lesions where defined using the union of all readers’ manual annotations. The less number of readers that agree on a lesion, the more difficult it can be considered. The CNN missed 2 lesions that all readers agreed on, and marked 7 regions that no reader marked

From: AI-based detection of lung lesions in [18F]FDG PET-CT from lung cancer patients

Number of readers agree Count Reader A Reader B Reader C CNN
3 18 18 18 18 16
2 1 1 1 0 1
1 9 1 5 3 3