小曲率半径医学图像快速三维重构算法研究

A Fast 3D Reconstruction Algorithm of Medical Image with Small Curvature Radius

针对小曲率半径医学器官图像表面不具有明显的特征,无法采用自由拼接,排序过程耗时严重,图像重构过程坐标失真、毛刺较多的问题。提出了一种医学图像曲面混乱区域配准方法,对小曲率半径图像三维坐标空间变换参数进行改进,在改进过程中采将坐标误差可接受程度收敛至最优值附近。摒弃传统的排序拼接思维。由后期需要融合的医学三维图像曲面分别进行小波分解及小波逆变换,重构得到高清晰医学图像三维曲面。试验结果表明:医学图像三维可视化绘制中,在不具备明显特征条件下,绘制时间缩短,融合后图像清晰度高。

Since there is no obvious characteristics on the surface of medical image with small curvature radius,problems will be encountered,including impossibility of image stitching,large time-consumption in image sequencing,coordinate distortion and formation of burr in image reconstruction process. To this end,a registration method for chaotic area of medical image surface was proposed to improve the 3 D coordinate space transformation parameter of medical image with small curvature radius. In improvement process,the acceptability of coordinate error was converged to the optimal value. The traditional thoughts on image sequencing and stitching were abandoned in this study,instead the 3 D medical images were successively subjected to wavelet decomposition and inverse wavelets transform before being fused,resulting in high-definition 3 D surface of medical image. Results showed that the 3 D visual drawing of medical image shortened the drawing time and improved the definition of post-fusion image even thought there is no obvious characteristics on the surface of medical image with small curvature radius.