中国组织工程研究

中国组织工程研究 ›› 2010, Vol. 14 ›› Issue (52): 9744-9747.doi: 10.3969/j.issn.1673-8225.2010. 52.014

• 骨与关节图像与影像 bone and joint imaging • 上一篇    下一篇

基于脑梗死大鼠MR图像熵的偏差场拟合方法

董  杰1，2，周  震3，刘春红2，王  影4，张  宇1   

  1. 1首都医科大学附属复兴医院放射科，北京市     100038；2首都医科大学附属安定医院放射科，北京市  100088；3首都医科大学生物医学工程学院，北京市   100069；4首都医科大学神经科学研究所，北京市  100069
  • 出版日期:2010-12-24 发布日期:2010-12-24
  • 通讯作者: 周震，博士，讲师，首都医科大学生物医学工程学院，北京市 100069 zhouzhenbme@gmail.com
  • 作者简介:董杰，男，1971年生，天津市人，汉族，北京大学医学部毕业，主治医师，主要从事算法设计和影像诊断。 dongjiedong1@163.com
  • 基金资助:

    首都医科大学基础-临床科研合作基金(10JL63)，课题名称：神经干细胞移植治疗脑梗塞大鼠模型的磁共振评价。

A bias field fitting method based on MR entropy for cerebral infarction rats

Dong Jie 1,2, Zhou Zhen3, Liu Chun-hong2, Wang Ying4, Zhang Yu1   

  1. 1 Department of Radiology, Fuxing Hospital of Capital Medical University, Beijing  100038, China; 2 Department of Radiology, Anding Hospital of Capital Medical University, Beijing  100088, China; 3 College of Biomedical Engineering, Capital Medical University, Beijing  100069, China; 4 Institute of Neuroscience, Capital Medical University, Beijing  100069, China
  • Online:2010-12-24 Published:2010-12-24
  • Contact: Zhou Zhen, Doctor, Lecturer, College of Biomedical Engineering, Capital Medical University, Beijing 100069, China zhouzhenbme@gmail.com
  • About author:Dong Jie, Attending physician, Department of Radiology, Fuxing Hospital of Capital Medical University, Beijing 100038, China; Department of Radiology, Anding Hospital of Capital Medical University, Beijing 100088, China dongjiedong1@163.com
  • Supported by:

    the Basic Clinical Scientific Research Program of Capital Medical University, No. 10JL63*

PDF

407

摘要：

背景：目前随着高场强MR设备的不断出现，由于受RF线圈等因素的影响，伪影成为MR图像不可避免的问题。在MR图像上呈现出无规律的、缓慢的灰度密度变化，相当于在增益场上叠加了一个偏差场。这不仅降低了MR图像的质量也严重干扰了数字图像分割和量化处理等方法的运用。
目的：提出一种以图像数据为基础并基于熵的复杂信号偏差场的修正方法。
方法：根据大鼠脑立体定位图谱，对脑梗死电凝模型大鼠进行冠状位MRI检查。采用GE Signa 1.5T磁共振扫描仪，3inch表面线圈。层厚6 mm，间隔1 mm(3D T2WI采用层厚2 mm，间隔0.5 mm)。该方法是校正场寻优的过程，优化过程以二维信息熵作为尺度。算法的目的是利用最优校正场，使带有偏差的MR图像去偏差，复原真实图像。
结果与结论：实验以真实和模拟数据对该方法进行实验，与传统方法相比，经该方法去偏差场后，其各组织在直方图中的分布和对比更加明显，更易于进行组织分割和组织提取。通过计算机时间比较，文章提出的方法在低密度Delaunay三角剖分下，比传统同态滤波方法省时；但在高密度Delaunay三角剖分下，比传统方法费时。

关键词: 熵, MR图像, 偏差场, 脑梗死, 伪影

Abstract:

BACKGROUND: In MR imaging, especially at high field strength, non-uniform RF coil response is a difficult problem to avoid. It is reflected as a smoothly but non-linearly varying bias field, which modulates tissue intensities across the acquired image. Such inhomogeneties degrade image quality in general and impede segmentation.
OBJECTIVE: To propose a model independent correction method of complex image signal bias field based on image data entropy.
METHODS: Data were collected by MR device (GE Signa 1.5T). MR images of the cerebral infarction rats were obtained (imaging slice thickness 6 mm, gap 1 mm), except for 3D T2WI (imaging slice thickness 2 mm, gap 0.5 mm). The method was to optimize and remove the bias field depending on 2D entropy.
RESULTS AND CONCLUSION: This method was compared to other methods on a number of simulated and real MR images. Compared with traditional methods, the tissues distribution and comparison were more obvious in the histogram, which ease segmentation and extraction. Compared with traditional methods, this method is more effective under low-density Delaunay triangulation, but costs more time under high-density Delaunay triangulation.

中图分类号: