分子影像学杂志2025，Vol.48Issue(6)：668-677,10.DOI:10.12122/j.issn.1674-4500.2025.06.02
人工智能辅助诊断系统与Lung-RADS对不同临床特征肺结节的良恶性预测效能
Efficacy of an artificial intelligence-assisted diagnostic system and Lung-RADS in predicting the benignity and malignancy of pulmonary nodules with different clinical characteristics
摘要
Abstract
Objective To evaluate the effectiveness of an artificial intelligence(AI)image-assisted diagnostic system in the prediction of pulmonary nodules and its clinical application value.Methods A total of 212 patients with definitive pathologic diagnoses of pulmonary nodules underwent analysis of their preoperative chest CT images,which were provided in DICOM format,using the AI-assisted diagnostic system.The diagnostic effectiveness of the AI model and Lung-RADS were compared in predicting of benign and malignant pulmonary nodules with different clinical and imaging characteristics.Results The AI model demonstrated higher diagnostic accuracy than Lung-RADS in distinguishing between benign and malignant pulmonary nodules(70.75%vs 60.85%,P<0.05).Results of the stratified analysis were as follows.By age:The AI model showed higher accuracy than Lung-RADS for patients aged 50-59 years(70.31%vs 53.13%,P<0.05).By nodule position:There were no significant differences between he AI model and Lung-RADS(P>0.05).By nodule density:The AI model showed higher accuracy than Lung-RADS for the mixed-ground glass nodules(74.51%vs 49.02%,P<0.05).By nodule size:The AI model showed higher accuracy than Lung-RADS for the nodules measuring 10-19 mm in diameter(74.75%vs 66.67%,P<0.05).By malignant pathology:The AI model exhibited higher accuracy in predicting adenocarcinoma nodules compared to Lung-RADS(77.52%vs 62.79%,P<0.05).Conclusion The AI image-assisted diagnostic system surpasses Lung-RADS in assessing the benign and malignant pulmonary nodules.With ongoing technological advancements,it has the potential to provide a reliable foundation for the early,non-invasive diagnosis of pulmonary nodules.关键词
肺结节/人工智能/影像辅助诊断系统/Lung-RADS/胸部CTKey words
lung nodules/artificial intelligence/image-assisted diagnostic system/Lung-RADS/chest CT引用本文复制引用
唐雅伦,李瑞,高磊,曹旸,乔炳礼,刘殿娜,姜敏,张毅鹏,胡凯文..人工智能辅助诊断系统与Lung-RADS对不同临床特征肺结节的良恶性预测效能[J].分子影像学杂志,2025,48(6):668-677,10.基金项目
国家自然科学基金面上项目(8217152484) （8217152484）
北京市科委课题(Z221100003522029) Supported by National Natural Science Foundation of China(8217152484). （Z221100003522029）
