机构地区: 广东工业大学计算机学院
出 处: 《中国生物医学工程学报》 2006年第6期672-677,共6页
摘 要: 本研究提出了一种自动识别颈部PET-CT图像特征点的算法,它应用自由变形(FFD)方法以CT图像的特征点为参考使PET图像产生变形,再结合最大互信息法对颈部PET与CT图像进行非刚体配准,最后用改进的小波图像融合法把两者进行融合得出视觉效果比较理想的融合图像。经实际计算得出的变形PET图像与对应CT图像的互信息量大于原始PET图像,并且最后用改进的小波图像融合法得出的融合图像的信息量比一般小波融合大,由此证明本研究所用方法是有效的。 An automatic recognition algorithm was developed to extract the feature points of PET and CT images. The free-form deformation (FFD) method was implemented to deform the PET image with feature points of PET and CT images. Then, the maximal mutual information method was used for nonrigid registration of the deformed PET image with CT image. Finally, the deformed PET and CT images were fused with improved wavelet fusion method. The mutual information was larger than the original by 0.7593% , and the entropy of the fused image with improved wavelet fusion method was larger than the normally produced image.
关 键 词: 自由变形 非刚体配准 小波融合 最大互信息法 医学图像
领 域: [自动化与计算机技术] [自动化与计算机技术]