基于相关性的三维医学图像插值

Interpolation of 3D Medical Images Based on Relativity

现有的插值方法在进行医学断层图像插值时,要么不能兼顾灰度和形状的变化,要么计算量太大。为解决这一问题,文中提出一种基于相关性的三维医学图像插值算法。通过对原图进行门限分割,获得体素的分割值。对相同密度物质采用灰度插值,不同密度物质,利用体素的相关性来进行插值,使新的图像不仅在灰度上,而且在组织形状上,介于原来的断层图像之间,满足了医学图像插值的要求。与线性插值相比,新算法的视觉效果好,计算误差小;插值结果可有效地应用于构建三维体模型。

In the case of medical image interpolation for 3D volume models,present methods either lack the capability of interpolating gray levels and shapes at the same time,or need higher computation cost. In order to solve the problem,this paper introduces a relativity-based 3D medical image interpolation algorithm. Firstly,the algorithm segments the original images with the threshold segmentation and obtains the segmentation value of voxels. Secondly,the algorithm uses gray-based interpolation in the same density matter,and makes use of the voxel relativity to interpolate image in the different density matter. The new image basically meets the requirements of medical image interpolation. Compared with linear interpolation,the algorithm proposed here greatly improves the quality of image. Moreover,the new algorithm has much lower computation cost compared with wavelet-based interpolation. The interpolation can be effectively used to construct 3D volume models.