多个人工智能辅助诊断系统对肺结节的诊断效能比较

Comparison of diagnostic efficacy of multiple artificial intelligence assisted diagnostic systems for pulmonary nodules

目的:探讨多个不同人工智能(AI)辅助诊断系统对肺结节的检出准确率及良恶性判断能力的差异。方法:选取中山大学附属第三医院岭南医院于2023年7月—2024年3月行肺结节切除手术的174例患者的胸部CT图像,将薄层肺算法图像数据传输至联影AI、深睿AI及数坤AI辅助诊断系统,以术后病理检查结果作为金标准,比较三个AI辅助诊疗系统对肺结节的诊断效能差异。结果:不同AI辅助诊断系统对肺结节的良恶性诊断准确率差异无统计学意义(P>0.05),对恶性肺结节检出的灵敏度、特异度及阳性预测值组间差异有统计学意义(P<0.05)。结论:AI辅助诊断系统在早期肺癌的检测中显示出较高的灵敏度,能够帮助放射科医生更高效地鉴别肺结节的良恶性,并一定程度上提高诊断的准确率,但目前不同AI辅助诊断系统对于肺结节良恶性判断的准确性仍有较大差异,在应用参考结果时需要慎重。

Objective To explore the difference of detection accuracy and benign and malignant judgment ability of multiple different artificial intelligence(AI)assisted diagnosis systems for pulmonary nodules.Methods The chest CT images of 174 patients who underwent pulmonary nodule resection in the Third Affiliated Hospital of Sun Yat-sen University Lingnan Hospital from July 2023 to March 2024 were selected.The thin-layer lung algorithm image data was transmitted to the UNITED IMAGING AI,Shen Rui AI and Shu Kun AI auxiliary diagnosis system.The postoperative pathological examination results were used as the gold standard to compare the diagnostic efficacy of the three AI-assisted diagnosis and treatment systems for pulmonary nodules.Results There was no significant difference in the diagnostic accuracy of benign and malignant pulmonary nodules between different AI-assisted diagnostic systems(P>0.05).There were significant differences in the sensitivity,specificity and positive predictive value of malignant pulmonary nodules(P<0.05).Conclusion AI-assisted diagnostic system shows high sensitivity in the detection of early lung cancer,which can help radiologists to identify benign and malignant pulmonary nodules more efficiently and improve the diagnostic accuracy to a certain extent.However,the accuracy of different AI-assisted diagnostic systems for the diagnosis of benign and malignant pulmonary nodules is still quite different,and it is necessary to be cautious when applying the reference results.