CT脑血管图像的三维重建图形配准算法

Registration Algorithm of CT Cerebrovascular Medical Image 3-D Reconstruction

在CT脑血管医学图像断层间三维重建图形配准中,提出了基于图像内部各个血管轮廓间相对位置信息的优化新算法。利用血管组织结构在系列断层切面上的重心位置具有连续性这一规律,在相邻两断层切片内寻找血管轮廓之间的配准关系。由于脑血管轮廓的形状比较对称,其重心与其几何中心相距甚微,因此,利用几何中心来代替重心,根据各轮廓的面积求其"质量",实现血管轮廓配准。实验结果表明,该方法具有全局最优和鲁棒性更强的特点。

A novel registration algorithm is presented for CT cerebrovascular medical image 3-D reconstruction based on the mutual location information of individual contours of the image set. This method is based on the rule that the barycenter of the individual slices from the same image series of body organ has continuity to match the contours of the neighboring slices. The shapes of the cerebrovascular contours are almost symmetric and the distance between barycenter and geometry center is small, So the barycenter can be replaced by the geometry center. The mass of the contour is calculated based on its area, then the overall ＂centroid＂ of the slice is found. The cerebrovascular contour registration is realized based on the facts above. Experimental results show that the algorithm has the characteristics such as global optimum and better robustness.