ISSN 1004-4140
CN 11-3017/P
苏寅晨, 张晓琴. 人工智能辅助诊断系统在肺结节检测及良恶性判断中的应用价值[J]. CT理论与应用研究. DOI: 10.15953/j.ctta.2023.128
引用本文: 苏寅晨, 张晓琴. 人工智能辅助诊断系统在肺结节检测及良恶性判断中的应用价值[J]. CT理论与应用研究. DOI: 10.15953/j.ctta.2023.128
SU Yinchen, ZHANG Xiaoqin. Application Value of Artificial Intelligence-assisted Diagnosis System in Lung Nodule Detection and Benign or Malignant Judgment[J]. CT Theory and Applications. DOI: 10.15953/j.ctta.2023.128
Citation: SU Yinchen, ZHANG Xiaoqin. Application Value of Artificial Intelligence-assisted Diagnosis System in Lung Nodule Detection and Benign or Malignant Judgment[J]. CT Theory and Applications. DOI: 10.15953/j.ctta.2023.128

人工智能辅助诊断系统在肺结节检测及良恶性判断中的应用价值

Application Value of Artificial Intelligence-assisted Diagnosis System in Lung Nodule Detection and Benign or Malignant Judgment

  • 摘要: 目的:探讨人工智能辅助诊断系统在肺结节的检测及良恶性判断中的应用价值。 方法:选择我院2022年3月至2023年3月行胸部CT检查,跟踪随访明确肺结节病理结果的患者作为研究对象,并制定真阳性结节标准。比较AI软件与影像医师对真阳性结节的检出价值及结节良恶性的判断价值。 结果:113名患者中AI软件检出1337个结节,医师检出774个结节,经验证共存在1079个真阳性结节。AI软件对于真阳性结节的检出率(98.98%)高于医师(71.27%),漏检率(1.02%)较医师(28.27%)低,误检率(23.91%)较医师(0.46%)高。AI软件对直径<5 mm及5 mm~10 mm真阳性结节的检出率(98.69%;100.00%)均高于医师(60.59%;80.25%);对于直径>10 mm真阳性结节的检出率(98.08%)稍高于医师(94.87%),但差异不具有统计学意义,需进一步选取临床大样本验证;AI软件对磨玻璃、实性、混合磨玻璃及钙化结节的检出率(98.47%;98.79%;100.00%;100.00%)均高于医师(75.52%;68.02%;72.73%;84.66%)。在结节良恶性判断方面,113名患者中共有115个结节经病理检查确诊,医师对结节良恶性判断的灵敏度(97.47%)、特异度(80.56%)、准确度(92.17%)均较AI(93.67%、66.67%、85.22%)高,但灵敏度和特异度的差异不具有统计学意义,可能与样本选择偏差有关;AI与病理结果对照取得了高度一致性(Kappa值为0.637),医师实现了几乎完全一致(Kappa值为0.811)。结论:AI辅助诊断对肺结节有较高的检出率,能大大降低漏检率,但误检率也随之上升;在肺结节良恶性鉴别中可为临床诊断提供辅助参考,但其准确性无法取代影像医师。

     

    Abstract: Objective: To investigate the application value of artificial intelligence-assisted diagnosis system in the detection and judgment of benign and malignant pulmonary nodules. Methods: Patients who underwent chest CT examination from March 2022 to March 2023 in our hospital and followed up with CT-guided needle biopsy or surgical procedures to determine the pathological nature were selected as the research objects, and the criteria for true positive nodules were formulated: On the basis of the image analysis results of AI software, two professional radiologists found suspicious lesions, and referred to multiplanar reconstruction and three-dimensional reconstruction and other images to determine the existence of lung nodules, and the two reports were consistently regarded as valid. The detection value of AI software and radiologists on the detection of true-positive nodules and the judgment value of benign and malignant nodules were compared. Results: AI software detected 1337 nodules in 113 patients; radiologists detected 774 nodules, and verified the coexistence of 1079 true-positive nodules. The detection rate of true positive nodules (98.98%) by AI software was higher than that of radiologists(71.27%), the missed diagnosis rate (1.02%) was lower than that of radiologists (28.27%), and the misdiagnosis rate (23.91%) was higher than that of radiologists (0.46%), and the above. The detection rate of <5 mm and 5mm-10mm true positive nodules with a diameter of 5mm and 5mm-10mm (98.69%; 100.00%)was higher than that of radiologists (60.59%; 80.25%); the detection rate of true positive nodules with a diameter of > 10 mm(98.08%)was slightly higher than that of radiologists (94.87%), but the difference was not statistically significant, and further clinical large samples were needed for verification. The detection rates of mixed ground-glass and calcified nodules (98.47%; 98.79%; 100.00%; 100.00%) were higher than those of radiologists (75.52%; 68.02%; 72.73%; 84.66%). In terms of the judgment of benign and malignant nodules, a total of 115 nodules in 113 patients were confirmed by pathological examination, of which 98 cases were obtained by interventional surgery or CT-guided biopsy in our hospital, and 17 cases were determined by follow-up diagnosis and treatment in other hospitals. The sensitivity (97.47%), specificity (80.56%), accuracy (92.17%) of the radiologists for the judgment of benign and malignant nodules were higher than those of AI (93.67%、66.67%、85.22%), However, the difference in sensitivity and specificity is not statistically significant and may be related to sample selection bias. AI achieved high agreement with pathological results (Kappa value of 0.637), and radiologists achieved almost complete agreement (Kappa value of 0.811). Conclusion: AI-assisted diagnosis has a high detection rate and sensitivity for pulmonary nodules, which can greatly reduce the missed diagnosis rate, but the misdiagnosis rate also increases. It can provide auxiliary reference for clinical diagnosis in the identification of benign and malignant pulmonary nodules, but its accuracy cannot replace that of radiologists.

     

/

返回文章
返回