CT人工智能技术检查指标联合miR-33a-5p鉴别诊断亚实性结节型肺腺癌浸润程度的价值

Value of CT Artificial Intelligence Technology Examination Indexes Combined with miR-33a-5p in Differential Diagnosis of the Invasive Degree of Solitary Subsolid Nodular Lung Adenocarcinoma

目的探讨CT人工智能技术检查指标联合miR-33a-5p鉴别诊断亚实性结节型肺腺癌浸润程度的价值。方法回顾性选取2021年4月—2023年2月首都医科大学大兴教学医院收治的98例亚实性结节型肺腺癌患者为研究对象。收集患者的临床资料,所有患者行胸部CT扫描,将扫描后的图像导入人工智能肺结节辅助诊断系统以完成自动检测和智能分析,采用PCR法检测肺组织miR-33a-5p相对表达量。根据肺组织病理检查结果将患者分为浸润组(38例)和微浸润组(60例)。采用多因素Logistic回归分析探讨亚实性结节型肺腺癌浸润程度的影响因素,采用ROC曲线分析CT人工智能技术检查指标、miR-33a-5p及其联合鉴别诊断亚实性结节型肺腺癌浸润程度的价值。结果浸润组病灶-肺界面清晰者占比、有分叶征者占比、有毛刺征者占比高于微浸润组,结节平均直径、平均CT值大于微浸润组,miR-33a-5p低于微浸润组(P<0.05)。多因素Logistic回归分析结果显示,病灶-肺界面、分叶征、毛刺征、结节平均直径、平均CT值是亚实性结节型肺腺癌浸润的独立危险因素,miR-33a-5p是亚实性结节型肺腺癌浸润的独立保护因素(P<0.05)。ROC曲线分析结果显示,病灶-肺界面、分叶征、毛刺征、结节平均直径、平均CT值、miR-33a-5p、联合检测鉴别诊断亚实性结节型肺腺癌浸润程度的AUC分别为0.654[95%CI(0.542~0.766)]、0.641[95%CI(0.528~0.754)]、0.650[95%CI(0.539~0.762)]、0.709[95%CI(0.602~0.816)]、0.670[95%CI(0.565~0.776)]、0.759[95%CI(0.666~0.852)]、0.935[95%CI(0.872~0.998)]。结论CT人工智能技术检查指标中的病灶-肺界面、分叶征、毛刺征、结节平均直径、平均CT值是亚实性结节型肺腺癌浸润的独立危险因素,而miR-33a-5p是亚实性结节型肺腺癌浸润的独立保护因素,且上述指标联合检测对亚实性结节型肺腺癌浸润程度的鉴别诊断价值较高。

Objective To investigate the value of CT artificial intelligence technology examination indexes combined with miR-33a-5p in differential diagnosis of the invasive degree of solitary subsolid nodular lung adenocarcinoma.Methods A total of 98 patients with solitary subsolid nodular lung adenocarcinoma admitted to Daxing Teaching Hospital,Capital Medical University from April 2021 to February 2023 were retrospective selected as the research subjects.The clinical data of the patients were collected,and all patients underwent chest CT scans,the scanned images were imported into the artificial intelligence-assisted pulmonary nodule diagnosis system for automatic detection and intelligent analysis.miR-33a-5p relative expression level was detected by PCR.Patients were divided into invasive group(n=38)and microinvasion group(n=60)based on the pathological results of lung tissue.Multivariate Logistic regression analysis was used to explore the influencing factors of invasive degree of solitary subsolid nodular lung adenocarcinoma.ROC curve was used to explore the value of CT artificial intelligence technology examination indexes,miR-33a-5p and their combined in differential diagnosis of the invasive degree of solitary subsolid nodular lung adenocarcinoma.Results The proportion of patients with clear lesion-lung interface,lobulation sign,spiculation sign in invasive group were higher than those in microinvasion group,average nodule diameter,and average CT value were larger than those in microinvasion group,and miR-33a-5p was lower than that in microinvasion group(P<0.05).Multivariate Logistic regression analysis showed that the lesion-lung interface,lobulation sign,spiculation sign,average nodule diameter,and average CT value were independent risk factors for invasive degree of solitary subsolid nodular lung adenocarcinoma,miR-33a-5p was a independent protective factor for invasive degree of solitary subsolid nodular lung adenocarcinoma(P<0.05).ROC curve analysis showed that the AUC of the lesion-lung interface,lobulation sign,spiculation sign,average nodule diameter,and average CT value and miR-33a-5p in differential diagnosis of the invasive degree of solitary subsolid nodular lung adenocarcinoma was 0.654[95%CI(0.542-0.766)],0.641[95%CI(0.528-0.754)],0.650[95%CI(0.539-0.762)],0.709[95%CI(0.602-0.816)],0.670[95%CI(0.565-0.776)],0.759[95%CI(0.666-0.852)],0.935[95%CI(0.872-0.998)],respectively.Conclusion Among CT artificial intelligence technology examination indexes,lesion-lung interface,lobulation sign,spiculation sign,average nodule diameter,and average CT value are independent risk factors for invasion of solitary subsolid nodular lung adenocarcinoma,miR-33a-5p is a independent protective factor for invasion of solitary subsolid nodular lung adenocarcinoma.Combination of above indexes has high value in differential diagnosis of the invasive degree of solitary subsolid nodular lung adenocarcinoma.