基于深度学习的CBCT下颌神经管自动分割算法综述  

Review of deep-learning based segmentation algorithms for mandibular nerve canal in CBCT images

作　　者：乐敏 李亮[1,2] LE Min;LI Liang(Department of Engineering Physics,Tsinghua University,Beijing 100084,China;Institute of Precision Medicine,Tsinghua University,Beijing 100084,China)

机构地区：[1]清华大学工程物理系,北京100084 [2]清华大学精准医学研究院,北京100084

出　　处：《中国体视学与图像分析》2023年第3期302-313,共12页Chinese Journal of Stereology and Image Analysis

基　　金：北京市自然科学基金-海淀原始创新联合基金(L222001)。

摘　　要：下颌神经管是非常重要的下颌骨结构,许多下颌手术都需要了解下颌神经管的位置和走向。目前,锥形束计算机断层成像(Cone beam computed tomography, CBCT)是主流的术前影像检测手段。但是人工从CBCT图像中识别、分割出下颌神经管是费时费力的一项工作,基于深度学习的自动分割技术成为当前的研究热点。本文对该领域的技术进展和研究现状进行了梳理,将近年来利用深度学习实现下颌神经管自动分割的算法进行了分类、汇总和比较,并讨论了该领域的未来发展方向。The mandibular canal is a very important mandibular structure.Many mandibular surgeries need to obtain the position and direction of the mandibular canal.At present,cone beam computed tomography(CBCT)is the mainstream preoperative imaging method.However,it is time-consuming and laborious to manually identify and segment a mandibular canal from CBCT images.Automatic segmentation technology based on deep learning has become a research hotspot currently.This paper systematically combs the technical progress and research status of this field.The algorithms for automatic segmentation of mandibular canals using deep learning in recent years are classified,summarized and compared.The future development direction of this field is also discussed.

关 键 词：深度学习 CBCT 下颌神经管 自动分割算法 

分 类 号：N1[自然科学总论] TP18[自动化与计算机技术—控制理论与控制工程] TP391.7[自动化与计算机技术—控制科学与工程]