利用三维自适应分裂-合并的MRI图像分割算法设计

为了在医学图像分割中,发现均匀几何三维区域的复杂形状,以提高分割准确率,提出一种基于3D几何特征分裂-合并(ASM)的脑部MRI图像分割算法；首先构建简单平行六面体的12种3D区域分割策略,体积分割技术将整个体积划分为许多大的均匀三维几何区；然后,在体积内定义更多小的均匀区域,以便在随后的合并步骤中有更大的生存概率；最后,进行多级区域合并,合并阶段只涉及复杂ASM树的叶子,考虑灰度相似性和共同边界区的大小,将小的区域合并为大邻近区；相比其他几种MRI图像分割算法,提出的方法在分割过程对噪声具有鲁棒性,提高了分割性能和准确率；另外提出的方法不需要训练数据集。

Design of MRI Image Segmentation Algorithm Using Three Dimensional Adaptive Split-Merge

In order to find the complex shape of the 3D uniform geometry in medical image segmentation and to improve the segmentation accuracy, a MRI image segmentation algorithm based on 3D geometry adaptive split-merge is proposed. Firstly, a simple parallel hexahedron 12 kinds of 3D region segmentation strategy is built, and volume segmentation technology is used to divide the whole volume into many large homogeneous 3D geometric regions. Then, more small homogeneous regions in the volume are defined, so as to have a greater probability of survival in the coming merge step. Finally, a multi-level region merging phase involves only complex ASM tree leaves, concerning gray and size of common boundary areas, merging small regions to adjacent areas. Compared with other MRI image segmentation algorithms, the proposed method is robust to noise in the segmentation process, which improves the performance and accuracy of segmentation. In addition, the proposed method does not require training data sets.