摘要
为了实现多模态医学图像的配准融合,提出一种加快寻优的医学图像互信息配准算法实现CT和MR图像的配准。该算法首先使用形态学方法提取图像的边界,再用力矩主轴法算出浮动图像进行刚性变换的初步平移量和旋转量,然后以此作为互信息法的初始参数进行寻优,找出最佳变换,实现CT和MR医学图像的自动刚性配准。该方法计算简单、运算量少。利用该配准算法实现融合的结果图像经过临床医生检验,认为达到临床诊断的要求,能辅助临床医生对疾病做出正确的诊断。
A new medical image mutual information registration method, which can speedup the optimized process is proposed for CT and MR medical image auto rigid Registration. This algorithm uses mathematic morphology method to extract the boundary of registration images at first, and then calculates the floating image' s centroid coordinates and the angles of rotation through principal axes algorithm. At last, the method uses these coordinates and angles as the initial parameters of the optimized process, then finds the best transformation values. This algorithm computes simply and with little time. The clinicians demonstrate that the registrated and fusion images of CT and MR images reach the need of clinical diagnosis, and can help diagnose more correctly.
出处
《中山大学学报(自然科学版)》
CAS
CSCD
北大核心
2005年第A02期174-177,共4页
Acta Scientiarum Naturalium Universitatis Sunyatseni
基金
广东省科技计划资助项目(2003B30602)
中山大学青年教师科研启动基金资助项目
关键词
力矩主轴
互信息
刚性配准
形态学
principal axes
mutual information
rigid registration
mathematic morphology