切换至 "中华医学电子期刊资源库"
高级检索
中华消化病与影像杂志(电子版)

中华消化病与影像杂志(电子版) ›› 2022, Vol. 12 ›› Issue (01) : 11 -15. doi: 10.3877/cma.j.issn.2095-2015.2022.01.003

论著

基于深度学习实现磁控胶囊内镜操作质量控制的可行性研究
沈大伟1, 李真1, 陈飞雪1, 张皓2, 彭程1, 孔庆周1, 刘静1, 田宝苓1, 李延青1, 任洪波1,†()   
  1. 1. 250012　济南，山东大学齐鲁医院消化内科
    2. 430000　安翰科技（武汉）股份有限公司
  • 收稿日期:2021-11-12 出版日期:2022-02-01
  • 通信作者: 任洪波
  • 基金资助:
    山东省重点研发计划（重大科技创新工程）(2019JZZY011007); 山东大学临床研究中心经费支持(2020SDUCRCC022)

A retrospective feasibility study of a quality-control system for real-time supervision of magnetically controlled capsule endoscopy based on deep learning

Dawei Shen1, Zhen Li1, Feixue Chen1, Hao Zhang2, Cheng Peng1, Qingzhou Kong1, Jing Liu1, Baoling Tian1, Yanqing Li1, Hongbo Ren1,†()   

  1. 1. Department of Gastroenterology, Qilu Hospital of Shandong University, Jinan 250012, China
    2. Ankon Technologies Co., Ltd, Wuhan 430000, China
  • Received:2021-11-12 Published:2022-02-01
  • Corresponding author: Hongbo Ren
全文 ( ) 下载PDF ( ) 导出引用
引用本文：

沈大伟, 李真, 陈飞雪, 张皓, 彭程, 孔庆周, 刘静, 田宝苓, 李延青, 任洪波. 基于深度学习实现磁控胶囊内镜操作质量控制的可行性研究[J]. 中华消化病与影像杂志(电子版), 2022, 12(01): 11-15.

Dawei Shen, Zhen Li, Feixue Chen, Hao Zhang, Cheng Peng, Qingzhou Kong, Jing Liu, Baoling Tian, Yanqing Li, Hongbo Ren. A retrospective feasibility study of a quality-control system for real-time supervision of magnetically controlled capsule endoscopy based on deep learning[J]. Chinese Journal of Digestion and Medical Imageology(Electronic Edition), 2022, 12(01): 11-15.

目的

开发一套基于深度学习的人工智能辅助质量控制系统，实现磁控胶囊胃镜检查过程实时质量控制。

方法

回顾性纳入山东大学齐鲁医院、山东美兆健康科技、莒县美年大健康三家内镜中心2019年1月至2020年1月共320例患者磁控胶囊胃镜检查图像资料，经内镜医师标注，基于Resnet-50开发出一套磁控胶囊胃镜辅助质量控制系统，实时进行检查完整度，部位清洁度，检查时间等指标监控，同时向操作者反馈检查质量，规范内镜操作者检查过程，实现磁控胶囊胃镜检查的质量控制。

结果

该磁控胶囊胃镜辅助质量控制系统识别胃内6解剖部位准确度为92.30~98.28%，4类清洁度识别准确度为91.46~95.37%，实际外部磁场操作时间计时准确度可达100%。同时，该质控系统整合至磁控操作系统，可安全运行。

结论

磁控胶囊胃镜辅助质量控制系统可实现质量控制，实时向内镜操作员反馈检查质量，规范内镜操作过程。

关键词: 磁控胶囊胃镜, 质量控制, 深度学习
Objective

To construct a real-time quality-control system during the stage of magnetically controlled capsule endoscopy for supervision based on deep learning.

Methods

A total of 320 patients with magnetically controlled capsule endoscopy image data were retrospectively included from January 2019 to January 2020 in endoscopy center of Qilu Hospital of Shandong University, Shandong Meizhao Healthy Technology and Juxian Meinian Onehealth Healthcare Holdings.After labeling by experienced endoscopists according to the guideline published in China, a real-time quality-control system including completeness with cleanliness and operation time report was developed based on the Resnet-50 model.With the help of real-time quality-control system, the operator can modulate the inspection process in terms of feedback information provided by quality-control system.

Results

The accuracy of a real-time quality-control system in identifying six anatomical parts of the gastric ranged from 92.30% to 98.28%, the accuracy of system in recognizing four types of mucosal cleanliness ranged from 91.46% to 95.37%, and the accuracy of timing of actual external magnetic field operation time was 100%.Meanwhile this system can be integrated into the magnetic control operating system safely and efficiently.

Conclusion

The real-time quality-control system of magnetically controlled capsule endoscopy can achieve the quality control function in the real-time inspection process, and provide real-time feedback to the endoscopic operator besides provide guidance and help for endoscopic operation.

Key words: Magnetically controlled capsule endoscopy, Quality control, Deep learning
表1 部位识别模型数据集分布(张)
部位 训练集图片数 测试集图片数
贲门 2 503 534
胃底 2 925 770
胃体 6 395 1 690
胃角 2 869 685
幽门 552 273
胃窦 7 121 1 617
合计 22 365 5 569
图1 人工智能辅助质量控制软件界面
图2 部位识别模型混淆矩阵 注：纵坐标为真实标签，横坐标为预测标签
图3 清洁度模型判断混淆矩阵 注：纵坐标为真实标签，横坐标为预测标签
表2 部位识别模型测试结果
部位 敏感度(95% CI) 特异度(95% CI) 阳性预测值(95% CI) 阴性预测值(95% CI) 准确度(95% CI)
贲门 0.8090 0.9728 0.7592 0.9796 0.9571
  0.7725 0.9678 0.7215 0.9752 0.9514
  0.8409 0.9770 0.7933 0.9833 0.9623
胃底 0.7247 0.9817 0.8638 0.9569 0.9461
  0.6914 0.9773 0.8343 0.9508 0.9399
  0.7557 0.9852 0.8888 0.9624 0.9519
胃体 0.8953 0.9350 0.8572 0.9535 0.9230
  0.8794 0.9267 0.8398 0.9462 0.9156
  0.9093 0.9425 0.8730 0.9598 0.9298
胃角 0.8526 0.9732 0.8168 0.9792 0.9583
  0.8233 0.9682 0.7860 0.9747 0.9528
  0.8778 0.9774 0.8441 0.9829 0.9634
幽门 0.8974 0.9872 0.7827 0.9947 0.9828
  0.8537 0.9837 0.7321 0.9922 0.9790
  0.9296 0.9899 0.8263 0.9964 0.9860
胃窦 0.8788 0.9646 0.9103 0.9511 0.9397
  0.8617 0.9582 0.8948 0.9438 0.9331
  0.8941 0.9700 0.9238 0.9575 0.9458
表3 清洁度识别模型测试结果
清洁度分类 敏感度(95% CI) 特异度(95% CI) 阳性预测值(95% CI) 阴性预测值(95% CI) 准确度(95% CI)
1分 0.9600 0.9516 0.8649 0.9866 0.9537
  0.9199 0.9308 0.8111 0.9727 0.9369
  0.9813 0.9666 0.9056 0.9938 0.9670
2分 0.7264 0.9803 0.9277 0.9113 0.9146
  0.6603 0.9773 0.8743 0.8862 0.8934
  0.7842 0.9852 0.9604 0.9314 0.9329
3分 0.7980 0.9579 0.8617 0.9351 0.9183
  0.7348 0.9380 0.8021 0.9123 0.8974
  0.8497 0.9717 0.9061 0.9525 0.9361
4分 0.9756 0.9285 0.8197 0.9913 0.9402
  0.9409 0.9044 0.7644 0.9786 0.9218
  0.9910 0.9470 0.8646 0.9968 0.9555
1
Jin G,，Lv J,，Yang M，et al.Genetic risk，incident gastric cancer，and healthy lifestyle：a meta-analysis of genome-wide association studies and prospective cohort study[J]．Lancet Oncol，2020，21(10)：1378-1386.
2
Necula L,，Matei L,，Dragu D，et al.Recent advances in gastric cancer early diagnosis[J]．World J Gastroenterol，2019，25(17)：2029-2044.
3
Yao K,，Uedo N,，Kamada T，et al.Guidelines for endoscopic diagnosis of early gastric cancer[J]．Dig Endosc，2020，32(5)：663-698.
4
廖专，李兆申．磁控胶囊胃镜：开启消化内镜新时代[J]．中华消化杂志，2019(06)：363-366.
5
Hu J,，Wang X,，Sun S．Comparison between the widely used magnetically controlled capsule gastroscopy and conventional gastroscopy：a meta-analysis[J]．Minim Invasive Ther Allied Technol，2021，1-14.
6
Geropoulos G,，Aquilina J,，Kakos C，et al.Magnetically Controlled Capsule Endoscopy Versus Conventional Gastroscopy：A Systematic Review and Meta-Analysis [J]．J Clin Gastroenterol，2021，55(7)：577-585.
7
上海国家消化系统疾病临床医学研究中心，国家消化内镜质控中心，中华医学会消化内镜学分会胶囊内镜协作组，等．中国磁控胶囊胃镜临床应用指南(精简版，2021年，上海)[J]．中华消化杂志，2021，41(9)：582-587.
8
Min J K,，Kwak M S,，Cha J M．Overview of Deep Learning in Gastrointestinal Endoscopy[J]．Gut Liver，2019，13(4)：388-393.
9
Horie Y,，Yoshio T,，Aoyama K，et al.Diagnostic outcomes of esophageal cancer by artificial intelligence using convolutional neural networks[J]．Gastrointest Endosc，2019，89(1)：25-32.
10
Ding Z,，Shi H,，Zhang H，et al.Gastroenterologist-Level Identification of Small-Bowel Diseases and Normal Variants by Capsule Endoscopy Using a Deep-Learning Model[J]．Gastroenterology，2019，157(4)：1044-1054.
11
Wu L,，Zhang J,，Zhou W，et al.Randomised controlled trial of WISENSE，a real-time quality improving system for monitoring blind spots during esophagogastroduodenoscopy[J]．Gut，2019，68(12)：2161-2169.
12
徐铭，姚理文，胡珊，等．基于深度学习的消化内镜检查辅助质量控制系统研究(含视频)[J]．中华消化内镜杂志，2021，38(2)：107-114.
13
Zhou J,，Wu L,，Wan X，et al.A novel artificial intelligence system for the assessment of bowel preparation(with video)[J]．Gastrointestinal Endoscopy，2020，91(2)：428-435.
14
Xia J,，Xia T,，Pan J，et al.Use of artificial intelligence for detection of gastric lesions by magnetically controlled capsule endoscopy[J]．Gastrointest Endosc，2021，93(1)：133-139.
15
廖专，王贵齐，陈刚，等．中国磁控胶囊胃镜临床应用专家共识(2017，上海)[J]．中华消化内镜杂志，2017，34(10)：685-694.
16
廖专，王贵齐，陈刚，等．中国磁控胶囊胃镜临床应用专家共识精简版(2017年，上海)[J]．中华消化杂志，2017，37(12)：793-795.
17
Zhu S,，Qian Y,，Tang X，et al.Gastric preparation for magnetically controlled capsule endoscopy：A prospective，randomized single-blinded controlled trial[J]．Dig Liver Dis，2018，50(1)：42-47.
18
He K,，Zhang X,，Ren S，et al.Deep residual learning for image recognition[C]//Proceedings of the IEEE conference on computer vision and pattern recognition.Nevada：USA，2016：770-778.
19
Ye S,，Lu S,，Bai X，et al.ResNet-Locust-BN Network-Based Automatic Identification of East Asian Migratory Locust Species and Instars from RGB Images[J]．Insects，2020，11(8)．
20
Beg S,，Card T,，Warburton S，et al.Diagnosis of Barrett′s esophagus and esophageal varices using a magnetically assisted capsule endoscopy system[J]．Gastrointestinal endoscopy，2020，91(4)：773-781.
21
Ching H,，Hale M F,，Sidhu R，et al.Magnetically assisted capsule endoscopy in suspected acute upper GI bleeding versus esophagogastroduodenoscopy in detecting focal lesions[J]．Gastrointestinal Endoscopy，2019，90(3)：430-439.
22
刘蓉，胡奉环，宋雷，等．磁控胶囊胃镜用于心血管疾病患者消化道检查的有效性及安全性[J]．中国循环杂志，2020，35(12)：1241-1244.
23
Kaan H L,，Phan P T,，Tiong A M H，et al.First-in-man feasibility study of a novel ingestible magnetically inflated balloon capsule for treatment of obesity[J]．Endoscopy International Open，2020，8(5)：E607-E610.
24
Xiao Y,，Wu Z,，He S，et al.Fully automated magnetically controlled capsule endoscopy for examination of the stomach and small bowel：a prospective，feasibility，two-centre study[J]．Lancet Gastroenterol Hepatol，2021，6(11)：914-921.
25
Rey J,，Ogata H,，Hosoe N，et al.Blinded nonrandomized comparative study of gastric examination with a magnetically guided capsule endoscope and standard videoendoscope[J]．Gastrointest Endosc，2012，75(2)：373-381.
26
Zou W,，Hou X,，Xin L，et al.Magnetic-controlled capsule endoscopy vs.gastroscopy for gastric diseases：a two-center self-controlled comparative trial[J]．Endoscopy，2015，47(6)：525.
27
Denzer U W,，Rosch T,，Hoytat B，et al.Magnetically guided capsule versus conventional gastroscopy for upper abdominal complaints：a prospective blinded study[J]．J Clin Gastroenterol，2015，49(2)：101-107.
28
Liao Z,，Hou X,，Lin-Hu E，et al.Accuracy of Magnetically Controlled Capsule Endoscopy，Compared With Conventional Gastroscopy，in Detection of Gastric Diseases[J]．Clin Gastroenterol Hepatol，2016，14(9)：1266-1273.
29
Hirasawa T,，Aoyama K,，Tanimoto T，et al.Application of artificial intelligence using a convolutional neural network for detecting gastric cancer in endoscopic images[J]．Gastric Cancer，2018，21(4)：653-660.
[1] 张梅芳, 谭莹, 朱巧珍, 温昕, 袁鹰, 秦越, 郭洪波, 侯伶秀, 黄文兰, 彭桂艳, 李胜利. 早孕期胎儿头臀长正中矢状切面超声图像的人工智能质控研究[J]. 中华医学超声杂志（电子版）, 2023, 20(09): 945-950.
[2] 陈舜, 薛恩生, 叶琴. PDCA在持续改进超声危急值管理制度中的价值[J]. 中华医学超声杂志（电子版）, 2023, 20(09): 974-978.
[3] 张丽丽, 陈莉, 余美琴, 聂小艳, 王婧玲, 刘婷. PDCA循环法在超声浅表器官亚专科建设中的应用[J]. 中华医学超声杂志（电子版）, 2023, 20(07): 717-721.
[4] 吕衡, 董理聪, 谢海琴, 赵卓非, 刘俐, 孙德胜. 基于CT-超声对照的肝脏局灶性病变超声漏诊状况分析：一项单中心横断面质量控制调查报告[J]. 中华医学超声杂志（电子版）, 2023, 20(07): 712-716.
[5] 钟露, 曹省, 宋宏宁, 陈金玲, 周青. 超声心动图定量评估二尖瓣反流程度的质量控制[J]. 中华医学超声杂志（电子版）, 2023, 20(07): 705-711.
[6] 武玺宁, 张培培, 齐振红, 张璟, 赵瑞娜, 孟华, 李建初. 腹膜后神经鞘瘤的超声诊断质量及其影响因素分析[J]. 中华医学超声杂志（电子版）, 2023, 20(07): 701-704.
[7] 方晔, 谢晓红, 罗辉. 品管圈在提高前列腺癌穿刺检出率中的应用[J]. 中华医学超声杂志（电子版）, 2023, 20(07): 722-727.
[8] 唐玮, 何融泉, 黄素宁. 深度学习在乳腺癌影像诊疗和预后预测中的应用[J]. 中华乳腺病杂志(电子版), 2023, 17(06): 323-328.
[9] 李锐颖, 危望, 王达志, 时志斌. 深度学习技术在膝关节疾病中的研究现状与展望[J]. 中华关节外科杂志(电子版), 2023, 17(05): 722-725.
[10] 姚宏伟, 魏鹏宇, 高加勒, 张忠涛. 不断提高腹腔镜右半结肠癌D3根治术的规范化[J]. 中华普外科手术学杂志(电子版), 2024, 18(01): 1-4.
[11] 李晓阳, 刘柏隆, 周祥福. 大数据及人工智能对女性盆底功能障碍性疾病的诊断及风险预测[J]. 中华腔镜泌尿外科杂志(电子版), 2023, 17(06): 549-552.
[12] 邢晓伟, 刘雨辰, 王明刚. 人工智能技术在疝和腹壁外科领域的应用及展望[J]. 中华疝和腹壁外科杂志(电子版), 2023, 17(04): 390-393.
[13] 雷漫诗, 邓锶锶, 汪昕蓉, 黄锦彬, 向青, 熊安妮, 孟占鳌. 人工智能辅助压缩感知技术在上腹部T2WI压脂序列中的应用[J]. 中华肝脏外科手术学电子杂志, 2023, 12(05): 551-556.
[14] 韩冰, 顾劲扬. 深度学习神经网络在肝癌诊疗中的研究及应用前景[J]. 中华肝脏外科手术学电子杂志, 2023, 12(05): 480-485.
[15] 胡平, 鄢腾峰, 周海柱, 祝新根. 人工智能在非增强CT图像中颅内出血早期检出和血肿分割的研究进展[J]. 中华脑血管病杂志(电子版), 2023, 17(04): 410-416.
阅读次数
全文


摘要

版权所有 中华医学会 中华医学电子音像出版社有限责任公司  京ICP备14006079号-1  网络出版服务许可证：(署)网出证(京)字第075号