基于人工智能技术的肺部CT扫描对肺纯磨玻璃结节浸润性病变及浸润程度的诊断价值  

Diagnostic value of pulmonary CT scanning based on artificial intelligence technology for the infiltrative lesions and degree of pulmonary ground glass nodules

作　　者：杨维伟[1] 吴明[1] 王宁[1] YANGWeiwei;WU Ming;WANG Ning(Department of Radiology,the First People's Hospital of Tianmen in Hubei Province,Tianmen,Hubei,431700,China)

机构地区：[1]天门市第一人民医院放射科,湖北天门431700

出　　处：《当代医学》2024年第5期79-83,共5页Contemporary Medicine

摘　　要：目的探讨基于人工智能技术的肺部CT扫描诊断肺纯磨玻璃结节浸润性病变及浸润程度的临床价值。方法回顾性分析2021年9月至2022年9月于天门市第一人民医院行手术治疗并经病理活检证实为肺纯磨玻璃结节的55例患者的临床资料,参考2021年世界卫生组织肺肿瘤新分类,根据病理结果分为非浸润性病变组(n=25,25个结节)与浸润性病变组(n=30,35个结节)。所有患者术前均行肺部CT检查,并利用人工智能系统对肺结节进行定性、定量参数分析。比较非浸润性病变组与浸润性病变组结节CT定性、定量参数;比较微浸润性腺癌、浸润性腺癌患者结节CT定性、定量参数;分析CT定性、定量参数对肺纯磨玻璃结节浸润程度预测的影响因素,并分析其预测效能。结果两组结节形态、分叶征、空泡征、支气管异常征、血管异常征、胸闷牵拉征占比及结节最长径、最短径、平扫CT值比较差异有统计学意义(P<0.05)。浸润性腺癌患者的结节形态规则占比低于微浸润性腺癌患者,分叶征、血管异常征、胸闷牵拉征占比均高于微浸润性腺癌患者,平扫CT值大于微浸润性腺癌患者,差异有统计学意义(P<0.05)。Logistic二元回归分析结果显示,分叶征、空泡征、血管异常征、平扫CT值增加是肺纯磨玻璃结节浸润性病变的危险因素(P<0.05);而分叶征、血管异常征、平扫CT值增加是肺纯磨玻璃结节浸润程度的危险因素(P<0.05)。ROC曲线分析结果显示,血管异常征诊断肺纯磨玻璃结节浸润性病变的效能最高(AUC为0.703,灵敏度92.00%,特异度48.60%),分叶征鉴别诊断肺纯磨玻璃结节浸润程度的效能最高(AUC为0.745,灵敏度75.00%,特异度73.90%)。结论基于人工智能技术的肺部CT扫描诊断肺纯磨玻璃结节浸润性及浸润程度有较高的临床价值,且人工智能技术CT扫描获得的内部血管征参数对诊断结节浸润性病变效能高,分叶征对鉴�Objective To explore the clinical value of pulmonary CT scanning based on artificial intelligence technology in diagnosing the infiltrative lesions and degree of pulmonary ground glass nodules.Methods The clinical data of 55 patients with pulmonary ground glass nodules confirmed by pathological biopsy after surgical treatment at the First People's Hospital of Tianmen in Hubei Province from September 2021 to September 2022 were retrospectively analyzed,the patients were divided into the non-invasive lesion group(n=25,25 nodules)and the invasive lesion group(n=30,35 nodules)according to the pathological results,reference the new classification of pulmonary tumors of the World Health Organization in 2021.All patients underwent pulmonary CT examination before operation,and artificial intelligence system was used to analyze the qualitative and quantitative parameters of pulmonary nodules.The qualitative and quantitative parameters of nodule CT were compared between the non-invasive lesion group and the invasive lesion group;the qualitative and quantitative parameters of nodule CT were compared between the patients with minimally invasive adenocarcinoma and invasive adenocarcinoma;the influence factors of qualitative and quantitative parameters of CT on the degree of infiltration of pure ground glass nodules of pulmonary were analyzed,and the prediction efficiency was analyzed.Results There were statistically significant differences in proportion of nodule morphology,lobulation sign,vacuole sign,bronchial anomaly sign,vascular anomaly sign,the chest tightness traction sign,and the longest diameter,the shortest diameter,and the plain CT value of nodules between the two groups(P<0.05).The proportion of regular nodule morphology in patients with invasive adenocarcinoma was lower than that in patients with minimally invasive adenocarcinoma,the proportion of lobulated sign,vascular abnormal sign and chest tightness traction sign in patients with invasive adenocarcinoma was significantly higher than those in patients with th

关 键 词：肺纯磨玻璃结节 肺部CT 人工智能技术 浸润性病变 浸润程度 

分 类 号：R734.2[医药卫生—肿瘤]