Relation Between Imaging Microfeatures of Artificial Intelligence-assisted Diagnosis System and Prognosis of Lung Adenocarcinomas Presented as Ground-glass Nodules

Objective To investigate the relation between the imaging microfeatures of AI-assisted diagnosis system and the prognosis of lung adenocarcinomas presented as ground-glass nodules (GGN).

Methods We retrospectively analyzed CT data of 162 patients with lung adenocarcinomas presented as GGN. According to different imaging characteristics, the patients were divided into pure ground glass nodules (PGGN) group and mixed ground glass nodules (MGGN) group. The AI-assisted diagnosis system was used to extract their imaging microfeatures, and their relation with the prognosis of the patients was analyzed.

Results The five-year OS and RFS were 89.7% and 88.5% in PGGN group, and 81.0% and 79.0% in MGGN group (χ²=6.289/7.255, P < 0.05). Multivariate Cox regression showed that imaging microfeatures such as microvascular cluster (P < 0.001), standard nodule volume (P=0.013) and nodule length (P < 0.001) were independent risk factors for OS, meanwhile, imaging microfeatures such as microvascular cluster (P < 0.001), standard nodule volume (P=0.017), nodule length (P=0.005), nodule central density (P=0.038) and lymph node metastasis (P < 0.001) were independent risk factors for RFS.

Conclusion The AI-assisted diagnosis system can effectively predict the prognosis of lung adenocarcinomas presented as GGN, and it also has a certain reference value for the clinical precision diagnosis and treatment of GGN and the prevention and treatment of early lung cancer.