作者：陆军平，邓少源，吴爱华，刘秋贤，李淑明

单位：佛山市南海区第四人民医院 消化内科， 广东 佛山 528200

Authors: Lu Junping，Deng Shaoyuan，Wu Aihua，Liu Qiuxian，Li Shuming

Unit: Department of Gastroenterology，The Fourth People's Hospital of Foshan Nanhai District，Foshan 528200，Guangdong，China

摘要：

目的 探究智能分光比色技术（FICE）在平坦型结肠腺瘤中的诊断价值及漏诊情况，分析漏诊的原因。方法 选择2018年5月至2022年5月佛山市南海区第四人民医院收治的100例平坦型结肠腺瘤患者作为研究对象，以病理诊断作为金标准，分析FICE的诊断效能及漏诊情况；多因素Logistic回归分析影响FICE诊断平坦型结肠腺瘤漏诊的因素，构建反向传播（BP）神经网络模型并评价其预测效能。结果 100例患者均经病理诊断为平坦型结肠腺瘤，Ⅱa型90例（90.00%），Ⅱb型8例（8.00%），Ⅱc型2例（2.00%），经FICE检查的敏感度为73.00%，检出率为75.00%,Kappa值为0.683，漏诊率为25.00%；多因素Logistic回归分析结果显示，年龄≥60岁、合并隆起型腺瘤、肠道准备不充分、腺瘤大小≤5 mm、腺瘤位置在右半结肠、医生经验<1000例、退镜时间<6 min是影响FICE诊断平坦型结肠腺瘤漏诊的危险因素（P<0.05）;BP神经网络模型的受试者操作特征曲线（ROC曲线）曲线下面积为0.873(95%CI:0.806～0.928,P<0.05)，敏感度和特异度分别为88.94%和82.46%，区分度较好，模型整体预测的准确性较高。结论 FICE在平坦型结肠腺瘤中具有较高的诊断价值，但仍存在漏诊情况，与患者的年龄、腺瘤大小、医生经验等因素有关，应引起临床重视，以提高诊断准确性。 

关键词：智能分光比色技术； 平坦型结肠腺瘤； 诊断效能； 漏诊； 危险因素

Abstract：

Objective To explore the diagnostic value and missed diagnosis of fuji intelligent chromoendoscopy (FICE) in flat colonic adenomas，and to analyze the causes of missed diagnosis. Method One hundred patients with flat colonic adenoma treated in our hospital from May 2018 to May 2022 were selected as the object of study. The diagnostic efficacy and missed diagnosis of FICE were analyzed with pathological diagnosis as the gold standard. Multivariate Logistic regression analysis was used to analyze the factors affecting the missed diagnosis of flat colonic adenoma by FICE，and to construct a back propagation (BP) neural network model and evaluate its predictive efficiency. Result Type Ⅱa 90 cases (90.00%)，type Ⅱb 8 cases (8.00%)，type Ⅱc 2 cases (2.00%). The sensitivity of FICE examination was 73.00%，the detection rate was 75.00% ，the Kappa value was 0.683，and the missed diagnosis rate was 25.00%. The results of multivariate Logistic regression analysis showed that age≥60 years old，complicated protuberant adenoma， inadequate intestinal preparation，adenoma size≤5 mm，adenoma location in the right colon，doctor's experience<1000 cases and mirror withdrawal time <6 min were the risk factors for missed diagnosis of flat colonic adenoma by FICE (P ＜0.05). The area under the receiver operating characteristic (ROC) curve of the BP neural network model is 0.873，the sensitivity and specificity are 88.94% and 82.46% respectively，and the overall prediction accuracy of the model is high. Conclusion FICE has high diagnostic value in flat colon adenoma，but it still has missed diagnosis，which is related to the age of patients，the size of adenoma， doctor's experience and other factors， which should be paid more attention to in order to improve the accuracy of diagnosis.

Key Words:  Fuji intelligent chromoendoscopy; Flat colonic adenoma; Diagnostic efficiency; Missed diagnosis; Risk factors