【摘要】 目的：探讨人工智能辅助诊断系统的应用对于提高在肺结节计算机体层成像（CT）检测准确性的影响。方法：选取2021年4月～2022年7月于本院接受胸部CT检查的309例患者作为研究对象，由影像科医师对CT影像进行审阅，应用人工智能辅助诊断系统进行阅片，结合人工阅片、应用人工智能辅助诊断系统阅片、人工阅片+应用人工智能辅助诊断系统阅片的结果，分析肺结核的检出情况，比较其差异性。结果：病理检查结果显示，在309例患者中，确诊肺结节1236个。人工阅片后，检出肺结节993个，存在漏检247个、误检4个。应用人工智能辅助诊断系统阅片后，检出肺结节1419个，存在漏检2个、误检185个。在肺结节CT检测中，人工阅片的漏检率高于应用人工智能辅助诊断系统阅片（P<0.05）。而人工智能辅助诊断系统阅片的误检率高于人工阅片（P<0.05）。在人工阅片的基础上，应用人工智能辅助诊断系统进行阅片，肺结节检出率为100%。在人工阅片+应用人工智能辅助诊断系统阅片的过程中，阅片时间为（299.73±79.78）s，比人工阅片更短（P<0.05）。应用人工智能辅助诊断系统阅片，直径<5mm、5～10mm、>10mm的肺结节的检出率分别为99.83%、100.00%、100.00%，实性结节、磨玻璃结节、部分实性结节的检出率分别为99.89%、99.66%、100.00%，显著高于人工阅片（P<0.05）。结论：在肺结节的临床诊断中实施胸部CT检查，影像科医师进行人工阅片的同时借助人工智能辅助诊断系统可以有效提高阅片效率，提高肺结节的检出率和准确性，降低误诊、漏诊的风险，便于及时、有效地进行治疗。更多还原

【Abstract】 Objective: To explore the influence of the application of artificial intelligence aided diagnosis system on the accuracy of CT detection of pulmonary nodules. Methods: 309 patients who received chest CT examination in our hospital from April 2021 to July 2022 were selected as the research objects. The CT images were reviewed by the radiologist, and the films were read by the AI aided diagnostic system. The detection of pulmonary tuberculosis was analyzed and the differences were compared based on the results of the manual film reading, the application of AI aided diagnostic system, and the application of AI assisted diagnostic system. Results: Pathological examination showed that 1236 pulmonary nodules were confirmed in 309 patients. After manual film reading, 993 pulmonary nodules were detected, 247 were missed and 4 were missed. After reading the films with the aid of artificial intelligence diagnostic system, 1419 pulmonary nodules were detected, 2 were missed and 185 were missed. In the CT detection of pulmonary nodules, the missed rate of manual film reading was higher than that of using artificial intelligence aided diagnostic system(P<0.05). However, the false detection rate of the artificial intelligence aided diagnosis system was higher than that of the manual reading(P<0.05). On the basis of manual film reading, the detection rate of pulmonary nodules was 100% when the artificial intelligence aided diagnosis system was used for film reading. In the process of manual film reading+application of artificial intelligence aided diagnosis system, the film reading time was(299.73±79.78)s, which was shorter than that of manual film reading(P<0.05). The detection rate of pulmonary nodules with diameters of <5mm, 5-10mm and >10mm was 99.83%, 100.00% and 100.00% respectively, and the detection rate of solid nodules, ground glass nodules and some solid nodules was 99.89%, 99.66% and 100.00% respectively, which was significantly higher than that of human workers(P<0.05). Conclusion: In the clinical diagnosis of pulmonary nodules, the implementation of chest CT examination and manual film reading by imaging physicians can effectively improve the film reading efficiency, improve the detection rate and accuracy of pulmonary nodules, reduce the risk of misdiagnosis and missed diagnosis, and facilitate timely and effective treatment.更多还原