Automatic anatomical site recognition of laryngoscopic images using convolutional neural network
-
摘要: 目的 探讨基于卷积神经网络(CNN)构建的人工智能(AI)质控系统对电子喉镜检查中的20个解剖站点的自动识别和分类情况。方法 回顾性收集中国医学科学院肿瘤医院内镜科2018年1月至12月电子喉镜检查的图像资料，采用Inception-ResNet-V2+SENet模型训练CNN。使用14 000张电子喉镜图像作为训练集，将这些图像分类到包含整个头颈部的20个具体解剖站点，并通过2000张喉镜图像和10个喉镜录像测试其性能。结果 训练后的CNN模型对每张喉镜图片识别的平均时间为(20.59±1.55) ms，对喉镜图像中20个解剖站点识别的总准确率为97.75%(1955/2000)，平均敏感性、特异性、阳性预测值和阴性预测值分别为100%、99.88%、97.76%和99.88%。该模型对喉镜录像中20个解剖站点识别的准确率≥99%。结论 基于CNN的AI系统可对电子喉镜图片及录像中的解剖部位进行准确、快速的分类识别，可用于喉镜检查中照片文档的质量控制，在监督喉镜检查质量方面表现出应用潜力。Abstract: Objective To explore the automatic recognition and classification of 20 anatomical sites in laryngoscopy by an artificial intelligence(AI) quality control system using convolutional neural network(CNN).Methods Laryngoscopic image data archived from laryngoscopy examinations at the Department of Endoscopy, Cancer Hospital, Chinese Academy of Medical Sciences from January to December 2018 were collected retrospectively, and a CNN model was constructed using Inception-ResNet-V2+SENet. Using 14000 electronic laryngoscope images as the training set, these images were classified into 20 specific anatomical sites including the whole head and neck, and their performance was tested by 2000 laryngoscope images and 10 laryngoscope videos.Results The average time of the trained CNN model for recognition of each laryngoscopic image was(20.59 ± 1.55) ms, and the overall accuracy of recognition of 20 anatomical sites in laryngoscopic images was 97.75%(1955/2000), with average sensitivity, specificity, positive predictive value, and negative predictive value of 100%, 99.88%, 97.76%, and 99.88%, respectively. The model had an accuracy of ≥ 99% for the identification of 20 anatomical sites in laryngoscopic videos.Conclusion This study confirms that the CNN-based AI system can perform accurate and fast classification and identification of anatomical sites in laryngoscopic pictures and videos, which can be used for quality control of photo documentation in laryngoscopy and shows potential application in monitoring the performance of laryngoscopy.
-
-
表 1 喉镜检查图像采集的20个具体解剖部位及各部位图片数量构成
主要解剖结构 具体图像采集站点 训练集(n) 测试集(n) 鼻腔 左侧鼻腔 700 100 右侧鼻腔 700 100 鼻咽 左侧鼻咽 700 100 右侧鼻咽 700 100 口咽 口咽及下咽(远景) 700 100 左侧咽会厌皱襞 700 100 右侧咽会厌皱襞 700 100 舌根和会厌谷 700 100 软腭(正中位) 700 100 左侧扁桃体 700 100 右侧扁桃体 700 100 下咽 下咽及喉部正中位(发衣音相) 700 100 下咽正中位(显露环后区) 700 100 左侧梨状窝 700 100 右侧梨状窝 700 100 喉部 喉部全景(吸气相) 700 100 双侧声带近景(吸气相) 700 100 口腔 口腔全景 700 100 硬腭 700 100 口底 700 100 合计 14 000 2000 
下载: 导出CSV
-
[1] 潘新良, 林云. 头颈部恶性肿瘤诊断与治疗的精准评估[J]. 中华耳鼻咽喉头颈外科杂志, 2022, 57(1): 89-95.
[2] 齐静怀, 张良. 人工智能时代的耳鼻咽喉头颈外科[J]. 临床耳鼻咽喉头颈外科杂志, 2020, 34(12): 1137-1140. https://lceh.cbpt.cnki.net/WKC/WebPublication/paperDigest.aspx?paperID=6464100d-37b2-4aee-a6e6-3263924d1f18
[3] Kröner PT, Engels MM, Glicksberg BS, et al. Artificial intelligence in gastroenterology: A state-of-the-art review[J]. World J Gastroenterol, 2021, 27(40): 6794-6824. doi: 10.3748/wjg.v27.i40.6794
[4] 付嘉, 李丽娟, 闫燕, 等. 深度学习辅助电子喉镜诊断喉白斑的应用研究[J]. 临床耳鼻咽喉头颈外科杂志, 2021, 35(5): 464-467. https://lceh.cbpt.cnki.net/WKC/WebPublication/paperDigest.aspx?paperID=59b2122a-0f33-4b15-8b1e-06f2109f953b
[5] 胡蓉, 钟琦, 徐文, 等. 基于深度卷积神经网络的人工智能在喉鳞状细胞癌窄带成像辅助诊断中的应用[J]. 中华耳鼻咽喉头颈外科杂志, 2021, 56(5): 454-458. doi: 10.3760/cma.j.cn115330-20200927-00773
[6] Xiong H, Lin P, Yu JG, et al. Computer-aided diagnosis of laryngeal cancer via deep learning based on laryngoscopic images[J]. EBioMedicine, 2019, 48: 92-99. doi: 10.1016/j.ebiom.2019.08.075
[7] Ren J, Jing X, Wang J, et al. Automatic Recognition of Laryngoscopic Images Using a Deep-Learning Technique[J]. Laryngoscope, 2020, 130(11): E686-E693.
[8] Paderno A, Piazza C, Del Bon F, et al. Deep Learning for Automatic Segmentation of Oral and Oropharyngeal Cancer Using Narrow Band Imaging: Preliminary Experience in a Clinical Perspective[J]. Front Oncol, 2021, 11: 626602. doi: 10.3389/fonc.2021.626602
[9] Hu J, Shen L, Albanie S, et al. Squeeze-and-Excitation Networks[J]. IEEE Trans Pattern Anal Mach Intell, 2020, 42(8): 2011-2023. doi: 10.1109/TPAMI.2019.2913372
[10] Kumai Y, Shono T, Waki K, et al. Detection of hypopharyngeal cancer(Tis, T1 and T2) by ENT physicians vs gastrointestinal endoscopists[J]. Auris Nasus Larynx, 2020, 47(1): 135-140. doi: 10.1016/j.anl.2019.05.007
[11] Sinonquel P, Eelbode T, Bossuyt P, et al. Artificial intelligence and its impact on quality improvement in upper and lower gastrointestinal endoscopy[J]. Dig Endosc, 2021, 33(2): 242-253. doi: 10.1111/den.13888
[12] Li YD, Zhu SW, Yu JP, et al. Intelligent detection endoscopic assistant: An artificial intelligence-based system for monitoring blind spots during esophagogastroduodenoscopy in real-time[J]. Dig Liver Dis, 2021, 53(2): 216-223. doi: 10.1016/j.dld.2020.11.017
[13] Wu L, He X, Liu M, et al. Evaluation of the effects of an artificial intelligence system on endoscopy quality and preliminary testing of its performance in detecting early gastric cancer: a randomized controlled trial[J]. Endoscopy, 2021, 53(12): 1199-1207. doi: 10.1055/a-1350-5583
[14] 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.
-
图(6)
表(1)
计量
- 文章访问数: 1530
- PDF下载数: 212
- 施引文献: 0