Fig. 6.

The demonstrations of the process of detecting different sizes of lung nodules by AI and the workflow diagram of the overall experimental design. (a): The four cases of different sizes of lung nodules with different characteristics and evolution show the importance of follow-up. (1): The upper lobe nodule (diameter: 2 mm, the green square frame) with tags to the pleural surface demonstrates some features of benign and stability at baseline; (2): A solid lung nodule (diameter: 4 mm, the green square frame) also demonstrates some features of benign and stable characteristics; (3, 4): The evolution of a small subsolid nodule in the right lung during the follow-up. (3): Inconspicuous small irregular nodule (diameter: 9 mm, the green square frame) adjacent to the right major fissure demonstrates acute margins to the fissure and does not satisfy criteria for an intrapulmonary lymph node. (4): Significant growth (diameter: 20 mm, the green square frame) is noted in the lesion approximately three months later due to progressive adenocarcinoma. (b): Workflow diagram showing the overall experimental design describing the flow of lung CT images through the labeling and grading process followed by creation of the IILS, which then underwent training and subsequent testing. The training dataset included images that passed sufficient quality standards from the clinical dataset. Subsequently, the output of the IILS was tested and compared with that of the traditional system. Finally, the impact of the process caused by the IILS was also assessed. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)