文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > Current methods in the automatic tissue segmentation of 3D magnetic resonance brain images,
侵权投诉

Current methods in the automatic tissue segmentation of 3D magnetic resonance brain images,

Abstract: Accurate segmentation of magnetic resonance (MR) images of the brain is of interest in the study of many brain disorders. In this paper, we provide a review of some of the current approaches in the tissue segmentation of MR brain images. We broad

Current Medical Imaging Reviews, 2006, 2, 000-0001

Current Methods in the Automatic Tissue Segmentation of 3D MagneticResonance Brain Images

Alan Wee-Chung Liew1 and Hong Yan2,3

12

Department of Computer Science and Engineering, The Chinese University of Hong Kong, Shatin, Hong Kong,Department of Computer Engineering and Information Technology, City University of Hong Kong, Kowloon Tong,Hong Kong, 3School of Electrical and Information Engineering, University of Sydney, NSW 2006, Australia

Abstract: Accurate segmentation of magnetic resonance (MR) images of the brain is of interest in the study of manybrain disorders. In this paper, we provide a review of some of the current approaches in the tissue segmentation of MRbrain images. We broadly divided current MR brain image segmentation algorithms into three categories: classification-based, region-based, and contour-based, and discuss the advantages and disadvantages of these approaches. We alsobriefly review our recent work in this area. We show that by incorporating two key ideas into the conventional fuzzy c-means clustering algorithm, we are able to take into account the local spatial context and compensate for the intensitynonuniformity (INU) artifact during the clustering process. We conclude this review by pointing to some possible futuredirections in this area.

Keywords: Medical imaging, magnetic resonance imaging, brain tissue segmentation, intensity nonuniformity artifact, partialvolume artifact, fuzzy clustering, image segmentation.1. INTRODUCTION

Magnetic resonance imaging (MRI) provides rich three-dimensional (3D) information about the human soft tissueanatomy [1]. It reveals fine details of anatomy, and yet isnoninvasive and does not require ionizing radiation such asx-rays. It is a highly flexible technique where contrastbetween one tissue and another in an image can be variedsimply by varying the way the image is made. For example,by altering radio-frequency (RF) and gradient pulses, and bycarefully choosing relaxation timings, it is possible tohighlight different components in the object being imagedand produce high contrast images. The rich anatomyinformation provided by MRI has made it an indispensabletool for medical diagnosis in recent years

Applications that use the morphologic contents of MRIfrequently require segmentation of the image volume intotissue types. For example, accurate segmentation of MRimages of the brain is of interest in the study of many braindisorders. In multiple sclerosis, quantification of whitematter lesions is necessary for drug treatment assessment [2].In schizophrenia, epilepsy or Alzheimer’s disease,volumetric analysis of gray matter (GM), white matter (WM)and cerebrospinal fluid (CSF) is important to characterizemorphological differences between subjects [3, 4, 5-8]. Suchstudies typically involve vast amount of data. Currently, inmany clinical studies segmentation is still mainly manual orstrongly supervised by a human expert. The level of operatorsupervision impacts the performance of the segmentationmethod in terms of time consumption, leading to infeasibleprocedures for large datasets. Manual segmentation alsoshows large intra- and inter-observer variability, making thesegmentation irreproducible and deteriorating the precision

*Address correspondence to this author at the Department of ComputerScience and Engineering, The Chinese University of Hong Kong, Shatin,Hong Kong; E-mail: wcliew@cse.cuhk.edu.hk; ityan@cityu.edu.hk

1573-4056/06 $50.00+.00

of the analysis of the segmentation. Hence, there is a realneed for automated MRI segmentation tools.

The automatic segmentation of MR images has been anarea of intense study [9, 10]. However, this task has provenproblematic, due to the many artifacts in the imagingprocess. In this paper, we review some of the currentapproaches in the tissue segmentation of MR brain images.We provide a mathematical formulation of the MRIsegmentation problem, and an overview of various MRIsegmentation methods, which we have broadly divided intothree categories: classification-based, region-based, andcontour-based. Then, we provide a discussion of our ownwork in this area, and provide some experimental resultsindicating the superior performance of our approach. Weconclude this review by pointing out some possible futuredirections.

2. MAGNETIC RESONANCE IMAGING

Magnetic resonance imaging is an imaging techniqueused primarily in medical settings to produce high qualityimages of the inside of the human body. In MRI, the imageis a map of the local transverse magnetization of thehydrogen nuclei. This transverse magnetization in turndepends on several intrinsic properties of the tissue. In thissection, we give a brief description of the principles of MRimaging, the readers are referred to [11, 12] for furtherdetails. MRI is based on the principles of nuclear magneticresonance (NMR). The NMR phenomenon relies on thefundamental property that protons and neutrons that make upa nucleus possess an intrinsic angular momentum calledspin. When protons and neutrons combine to form nucleus,they combine with oppositely oriented spins. Thus, nucleiwith an even number of protons and neutrons have no netspin, whereas nuclei with an odd number of protons orneutrons possess a net spin. Hydrogen nuclei have an NMRsignal since its nucleus is made up of only a single proton

©2006 Bentham Science Publishers Ltd.

第1页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email