摘要
提出了基于互信息的EAA(Extraction of Available Area有效区域提取)算法的配准方法。此方法根据人体脑部医学图像特点,首先通过图像的灰度差异,对图像进行预处理,并利用区域生长法,提取互信息的有效区域——病灶的可疑区域和颅骨轮廓,然后只将此区域做为配准的有效信息,寻找配准参数,使两幅图像的互信息最大。仿真时采用MR-PET图像,进行了22组对比实验。结果表明,此方法一定程度上消除了图像中无效区域的影响,在配准精度及配准时间上有一定的优势。
This paper proposes an algorithm of medical image registration based on extraction ot available area and maximal mutual information. At first, using the characteristics of human brain medical images, and the pretreatment of the images according to the grey differences, the algorithm extracts the available areas related to the information of the conceivable focuses and the skull by a region growing algorithm. At last, maximal mutual information on the areas is used in registration of images. Experiments on 22 contrastive tests of the MR- PET images demonstrated the effectiveness of this algorithm. The results show that this algorithm can reduces the influence of unavailable areas and it has the advantage of precision and computed time.
出处
《青岛大学学报（自然科学版）》
CAS
2007年第3期54-59,共6页
Journal of Qingdao University(Natural Science Edition)
关键词
有效区域提取
最大互信息
配准
脑部医学图像
human brain medical images
EAA
extraction of available area
the maximal mutual information
registration
