AI定量参数联合CT影像特征对肺占位性病变的诊断价值  

Diagnostic value of AI quantitative parameters combined with CT imaging features in pulmonary space-occupying lesions

作　　者：彭世芳 王英超 张曼 陈伟彬 PENG Shifang;WANG Yingchao;ZHANG Man;CHEN Weibin(CT Department,The Affiliated Hospital of North China University of Science and Technology,Tangshan 063000,China)

机构地区：[1]华北理工大学附属医院CT室,河北唐山063000

出　　处：《华北理工大学学报（医学版）》2024年第6期421-426,共6页Journal of North China University of Science and Technology：Health Sciences Edition

基　　金：2023年政府资助临床医学优秀人才培养项目(编号:ZF2023009)。

摘　　要：目的探讨AI定量参数、CT影像特征对肺占位性病变良恶性诊断及肺腺癌病变浸润性的评估价值,为临床诊治提供数据支持。方法收集2023年1月-2024年4月于华北理工大学附属医院接受胸腔镜手术治疗的175例肺占位性病变患者,根据术后病理结果分为良性病变组、恶性病变组,分析两组的AI定量参数、CT影像特征;进一步对恶性病变组根据浸润与否分为浸润性病变组(>0.5 cm)、非浸润性病变组(<0.5cm),比较两组参数间的差异,以ROC曲线分析各参数对肺占位性病变良恶性和肺腺癌是否浸润的评估价值。结果3D长径、分叶征、毛刺征、实性占比、CT平均值均为肺占位性病变良恶性的独立预测因子(P<0.05),联合参数评估肺占位性病变良恶性的AUC为0.884。3D长径、毛刺征、实性占比、CT最大值为肺腺癌病变浸润性的独立预测因子(P<0.05),联合参数评估浸润性的AUC为0.899。结论AI定量参数联合CT影像特征可以较好地评估肺占位性病变良恶性及肺腺癌浸润与否,且AI定量参数能为患者提供多数据支持。Objective To explore the diagnostic value of AI quantitative parameters and CT imaging features in differentiating benign and malignant pulmonary space-occupying lesions and assessing the invasiveness of lung adenocarcinoma,providing data support for clinical diagnosis and treatment.Methods A total of 175 patients with pulmonary space-occupying lesions who underwent thoracoscopic surgery at the Affiliated Hospital of North China University of Science and Technology between January 2023 and April 2024 were enrolled.Based on postoperative pathological results,the patients were divided into benign lesion group and malignant lesion group.AI quantitative parameters and CT imaging features were analyzed in both groups.Further,the malignant lesion group was subdivided into invasive lesion group(>0.5 cm)and non-invasive lesion group(<0.5 cm)based on invasion status.Differences in parameters between the groups were compared,and ROC curves were used to analyze the diagnostic value of each parameter in differentiating benign and malignant pulmonary space-occupying lesions and assessing the invasiveness of lung adenocarcinoma.Results 3D maximum diameter,lobulated margin,spiculated margin,solid component ratio,and mean CT value were independent predictors of benign and malignant pulmonary space-occupying lesions(P<0.05).The AUC for combined parameters in differentiating benign and malignant lesions was 0.884.3D maximum diameter,spiculated margin,solid component ratio,and maximum CT value were independent predictors of the invasiveness of lung adenocarcinoma(P<0.05).The AUC for combined parameters in assessing invasiveness was 0.899.Conclusion AI quantitative parameters combined with CT imaging features can effectively evaluate the benignity and malignancy of pulmonary space-occupying lesions and the invasiveness of lung adenocarcinoma.Moreover,AI quantitative parameters provide multiple data support for patients.

关 键 词：肺占位性病变 人工智能 CT影像特征 肺腺癌 浸润程度 鉴别价值 

分 类 号：R816.4[医药卫生—放射医学] R563[医药卫生—临床医学]