图像型全自动骨龄评测算法及应用研究进展  

Image based automatic bone age assessment algorithms and applications

作　　者：贾阳 陈伟光[1,2,3] 王海娟 杜涵容 韩俊岗 靳晓辉 杨斌 JIA Yang;CHEN Weiguang;WANG Haijuan;DU Hanrong;HAN Jungang;JIN Xiaohui;YANG Bin(School of Computer Science and Technology,Xi'an University of Posts and Telecommunications,Xi'an 710121,China;Shaanxi Key Laboratory of Network Data Intelligent Processing,Xi'an University of Posts and Telecommunications,Xi'an 710121,China;Xi'an Key Laboratory of Big Data and Intelligent Computing,Xi'an University of Posts and Telecommunications,Xi'an 710121,China;Department of Radiology,Xi'an Honghui Hospital,Xi'an 710054,China)

机构地区：[1]西安邮电大学计算机学院,陕西西安710121 [2]西安邮电大学陕西省网络数据智能处理重点实验室,陕西西安710121 [3]西安邮电大学西安市大数据与智能计算重点实验室,陕西西安710121 [4]西安市红会医院放射科,陕西西安710054

出　　处：《西安邮电大学学报》2021年第4期65-78,共14页Journal of Xi’an University of Posts and Telecommunications

基　　金：陕西省重点研发计划项目(2019GY-021);陕西省教育厅自然科学研究计划项目(18JK0722)。

摘　　要：通过分析图像型全自动骨龄评测算法所依赖的医学基础理论,对比图像型全自动骨龄评测中必要的图像分割方法和特征预测模型。归纳总结基于深度学习的自动骨龄评测算法,包括各类深度神经网络模型、模型训练所使用的数据集和骨龄评测结果等。此外,介绍骨龄评测中使用的其他数据采集方法及图像型全自动骨龄评测的发展情况,分析常见的几种全自动骨龄评测软件及系统,并对目前图像型全自动骨龄评测中存在的问题进行讨论与展望。Through the analysis of the basic medical theory of image-based automatic bone age assessment(BAA)algorithms,the image segmentation methods and feature prediction models in BAA are compared.The methods of BAA based on deep learning are summarized,including deep neural network models,data set used by model training,and BAA results.In addition,other methods adopted in BAA and the development of image-based automatic BAA are also summarized.Several common automatic BAA software and systems are analyzed.Existing problems of automatic BAA and its future development are discussed.

关 键 词：全自动骨龄评测 深度学习 手骨分割方法 X线片 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]