基于移动最小二乘法的医学图像配准  

Medical Image Registration Based on Moving Least Squares

作　　者：王伟[1] 苏志勋[1] 

机构地区：[1]大连理工大学数学科学学院,大连116024

出　　处：《计算机科学》2010年第9期270-271,293,共3页Computer Science

基　　金：国家自然科学基金(No.60673006)资助

摘　　要：提出一种基于移动最小二乘法变形模型的医学图像配准技术。首先用蛇模型的方法分割图像感兴趣区域;其次在分割后的图像上半自动地选取对应标记点;最后基于这些标记点采用移动最小二乘法的变形模型对图像进行变形,从而实现医学图像的配准。实验结果表明,该方法克服了手动选点难度大的缺点,提高了配准的精度,是一种有效的医学图像配准方法。Abstract The purpose of medical image registration is to match corresponding points in spatial and anatomic location from two or more different images.A novel algorithm of medical image registration based on moving least squares was proposed.This algorithm segmented interested regions of images,and then used semi-automatic method to extract the landmark points,deformed the image using moving least squares,thereby achieving medical image registration.The experimental results demonstrate that the proposed algorithm is an accuracy and effective method of medical image registration.

关 键 词：图像分割 图像配准 移动最小二乘法 半自动选点 蛇模型 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]