分析肺部影像人工智能诊断系统鉴别诊断肺结节性质

Differential diagnosis of pulmonary nodules and pathological types of lung cancer by pulmonary imaging artificial intelligence diagnostic system

目的探讨肺部影像人工智能诊断系统对肺结节性质及肺癌病理类型的鉴别诊断价值。方法回顾性分析医院2018年3月~2019年3月收治的178例肺结节患者的临床资料,经肺部影像人工智能诊断系统鉴别肺结节性质及肺癌病理类型,所有患者均经手术治疗,肺结节均经病理检查明确良恶性,恶性结节也均明确病理类型。统计肺部影像人工智能诊断系统的鉴别诊断结果;分析该方法与病理检查结果的一致性。结果本组患者中恶性结节占比38.78%,肺癌患者占比17.98%,且腺癌、小细胞癌、鳞癌、大细胞癌占比分别为71.88%、18.75%、6.25%、3.13%;经肺部影像人工智能诊断系统鉴别有152个良性肺结节,有93个恶性肺结节,有148例患者被诊断为良性肺结节病,有30例患者被诊断为肺癌,其中有21例被诊断为腺癌,有6例被诊断为小细胞癌,有2例被诊断为鳞癌,有1例被诊断为大细胞癌;经Kappa一致性检验,肺部影像人工智能诊断系统鉴别肺结节良恶性、肺癌病理类型结果与病理检查结果的一致性良好(Kappa值=0.801,P=0.015;Kappa值=0.763,P=0.024)。结论肺部影像人工智能诊断系统鉴别诊断肺结节性质及肺癌病理类型结果与病理检查结果一致性均较高,效能良好,值得推广。

Objective To explore the value of pulmonary imaging artificial intelligence diagnostic system in differential diagnosis of pulmonary nodules and pathological types of lung cancer.Methods The clinical data of 178 patients with pulmonary nodules admitted to our hospital from March 2018 to March 2019 were retrospectively analyzed.All of them were identified by pulmonary imaging artificial intelligence diagnostic system.All patients were treated surgically.The benign and malignant pulmonary nodules were confirmed by pathological examination,and the malignant nodules were also identified by pathological type.The results of differential diagnosis of AI diagnostic system of lung imaging were counted.The consistency between this method and pathological examination results were analyzed.Results In this group,malignant nodules accounted for 38.78%,lung cancer for 17.98%,and adenocarcinoma,small cell cancer,squamous cell cancer and large cell cancer accounted for 71.88%,18.75%,6.25%,3.13%,respectively.There were 152 benign pulmonary nodules,93 malignant pulmonary nodules,148 cases of benign pulmonary sarcoidosis,30 cases of lung cancer,including 21 cases of adenocarcinoma,6 cases of small cell carcinoma,2 cases of squamous cell carcinoma and 1 case of large cell carcinoma.The consistency test of Kappa showed that the AI diagnostic system of lung imaging was in good agreement with the pathological results of lung cancer(Kappa value=0.801,P=0.015;Kappa value=0.763,P=0.024).Conclusion The pulmonary imaging artificial intelligence diagnostic system has high consistency with pathological examination results in differential diagnosis of pulmonary nodules and pathological types of lung cancer,with good efficacy,which is worth widely popularizing.