Automatic Localization of Vertebrae Based on Convolutional Neural Networks

doi:10.1117/12.2081941

CORC > 自动化研究所 > 中国科学院自动化研究所 > 中国科学院分子影像重点实验室

	Automatic Localization of Vertebrae Based on Convolutional Neural Networks
	Shen W(沈伟)1 ; Yang F(杨凤)2; Mou W(牟玮)1 ; Yang CY(杨彩云)1 ; Tian J(田捷)1 ; Tian J(田捷)
	2015
会议日期	2015-2-21
会议地点	Orlando
关键词	Vertebrae Localization Hand-crafted Features Convolutional Neural Networks
DOI	10.1117/12.2081941
英文摘要	Localization of the spine is of importance in many medical applications. For example, the spine can serve as the landmark in image registration. It can also provide a reference coordinate system to facilitate the localization of other organs in the chest. Since the localization of the spine is based on that of the vertebrae which compose the spine, the accuracy of the vertebrae localization becomes very critical. In this paper, we propose a new vertebrae localization method using convolutional neural networks (CNNs). The main advantage of the proposed method is the removal of hand-crafted features. Firstly, we construct two training sets to train two CNNs that share the same architecture. One is used to distinguish the vertebrae from other structures in the chest, and the other is aimed at detecting the centers of the vertebrae. The architecture contains two convolutional layers, either of which is followed by a max-pooling layer. Then the output feature vector from the max-pooling layer is fed into a multilayer perceptron (MLP) classifier which has one hidden layer. Experiments were performed on ten chest CT images. We used leave-one-out strategy to train and test the proposed method. Quantitative comparison between the predict centers and ground truth shows that our convolutional neural networks can achieve promising localization accuracy without hand-crafted features.
会议录	SPIE
学科主题	医学图像处理
内容类型	会议论文
源URL	[http://ir.ia.ac.cn/handle/173211/11698]
专题	自动化研究所_中国科学院分子影像重点实验室
通讯作者	Tian J(田捷)
作者单位	1.Intelligent Medical Research Center, Institute of Automation, Chinese Academy of Sciences 2.School of Computer and Information Technology, Beijing Jiaotong University
推荐引用方式 GB/T 7714	Shen W,Yang F,Mou W,et al. Automatic Localization of Vertebrae Based on Convolutional Neural Networks[C]. 见:. Orlando. 2015-2-21.