摘要
随着计算机技术的不断发展和进步,深度学习在临床医学影像领域的辅助诊断方面取得了突破性进展。近年来,深度学习也被成功引入口腔医学领域,尤其以卷积神经网络(Convolutional Neural Networks,CNN)为代表的经典模型展示出强大的鲁棒性和普适性,即推动口腔医学走向数字化、智能化、自动化,从而实现数据合作共享。深度学习在口腔常见疾病的辅助诊断、口腔医学影像中解剖结构的识别定位与分割、指导口腔医师精细的临床操作等方面都有较成功的案例。相较之前传统手工操作及医师诊断的精准度均有所提高。然而深度学习技术在口腔医学领域的研究及应用正处于萌芽阶段,有待进一步的经验积累及检测。本文将深度学习与CNN的特点进行了阐述;总结出CNN在国内外口腔医学影像领域的应用现状;讨论及归纳了深度学习方法目前应用于口腔医学的问题及展望应用前景。
With the continuous development and progress of computer technology,Deep Learning has made a breakthrough in assisting diagnosis in the field of clinical medical imaging.In recent years,Deep Learning has also been successfully introduced into the field of,stomatology.In particular,the classical models represented by Convolutional Neural Networks(CNN)show strong robustness and universality,which promotes stomatology toward digitalization,intelligence and automation,so as to realize data cooperation and sharing.Deep Learning has more successful cases in aiding the diagnosis of common diseases in dentistry,identifying,positioning and segmentation of anatomical structures in dentistry images,and guiding dentists in fine clinical operations.The accuracy of traditional manual operation and diagnosis has been improved compared with the previous ones.However,the research and application of Deep Learning technology in the field of stomatology is still in its infancy,and further experience and testing are needed.In this paper,we firstly explain the characteristics of Deep Learning and Convolutional Neural Network;Ssecondly,the application status of CNN in the field of stomatological imaging at home and abroad was summarized.Finally,the problems and prospects of deep learning in stomatology are discussed and summarized.
作者
程一彤
武峰
CHENG Yi-tong;WU Feng(Dept.of Prosthodontics,Shanxi Medical University School and Hospital of Stomatology,Taiyuan 030001,China)
出处
《中华老年口腔医学杂志》
2022年第4期233-238,共6页
Chinese Journal of Geriatric Dentistry
基金
山西省重点研发计划项目(项目编号:201903D321120)。
关键词
口腔医学
深度学习
卷积神经网络
影像诊断
Stomatology
Deep learning
Convolutional Neural Network
Image diagnosis
