一种小波域上运用变分模型的PET/CT图像融合方法  

Variational Model on Wavelet Domain for PET/CT Image Fusion

作　　者：倪倩[1] 陈超敏[1] 周凌宏[1] 

机构地区：[1]南方医科大学生物医学工程学院,广州510515

出　　处：《医疗卫生装备》2008年第5期8-10,共3页Chinese Medical Equipment Journal

基　　金：广东省自然基金资助项目(06024583);广东省科技计划资助项目(2007B010400056)

摘　　要：目的:研究一种新的算法用于PET/CT图像的异机融合。方法:在小波域上运用变分的方法,结合小波和变分的优点,首先对PET和CT图像进行小波分解,然后运用提出的两个假设分别对分解后的近似和细节分量进行融合,最后进行小波逆变换得到融合图像。结果:根据具体的融合算法实例,并与MATLAB的小波融合工具箱的融合结果进行了比较,发现新算法对图像中病变区域的解剖定位更加的准确。结论:实验证明在小波域上运用变分法进行图像的融合可以得到很好的效果,能够更好地满足临床治疗的需要。Objective To find a new algorithm for PET/CT image fusion. Methods A variational model was used based on the wavelet transform, Firstly, PET and CT images were decomposed using wavelet transform, Then, images in approximate channel and detail channel were fused according to the two proposed assumption, Finally, decomposed images were synthesized to form fused image, Results Compared with the results form MATLAB wavelet fusion toolbox, the experimental results showed that the new variational image fusion model could provide more accurate result for target location in radiotherapy planning. Conclusion According to experiments, the new algorithm can reach good results and meet requirement of clinical demands.

关 键 词：图像融合 变分方法 小波 PET CT 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]