基于深度学习的全景片中颌骨疾病分类研究  

Research on Classification of Jaw Lesion in Panoramic X-Ray Images Based on Deep Learning Algorithm

作　　者：任家银[1] 郭际香[2] REN Jia-yin;GUO Ji-xiang(West China School of Stomatology,Sichuan University,Chengdu 610065;College of Computer Science,Sichuan University,Chengdu 610065)

机构地区：[1]四川大学华西口腔医学院,成都610065 [2]四川大学计算机学院,成都610065

出　　处：《现代计算机》2020年第14期35-38,共4页Modern Computer

基　　金：四川大学华西口腔医院青年科学研究基金(No.WCHS-201608)。

摘　　要：作为当前人工智能领域的主流技术,深度学习技术在医学图像处理如分割、检测、分类及识别领域取得成功的应用。颌骨疾病是一种人类常见、多发的口腔疾病,口腔全景片是其主要检查方式之一。及时从全景片中检测颌骨疾病,有助于提高医生工作效率,能为病情诊断提供一定的参考。提出一种基于深度学习的颌骨疾病四分类模型,能识别和分类正常全景片和三种常见的颌骨疾病。实验结果较好,四分类模型在囊肿病变的精确率达90%。As a promising method in artificial intelligence,deep learning has obtained many successful applications in medical images processing,such as segmentation,detection,classification and recognition.Jaw disease is a common and frequently encountered oral disease.Timely detection of jaw lesion from the orthopanotomography image is helpful to improve the work efficiency and reduce the workload of medical staffs,and can provide more reference information for disease diagnosis.A deep learning based model for jaw lesions classification was pro⁃posed,which can identify and classify the normal case and three jaw diseases.The good experimental result was obtained,the precision of four-classification on cysts disease is 90%.

关 键 词：卷积神经网络 口腔全景片 颌骨疾病 分类 

分 类 号：R782[医药卫生—口腔医学] TP18[医药卫生—临床医学] TP391.41[自动化与计算机技术—控制理论与控制工程]