结合形状特征点和灰度信息的心脏DSCT图像配准

Nonrigid Registration of Cardiac DSCT Images by Integrating Intensity and Point-feature

论文引进Marching Cubes算法提取心脏腔室结构形状标记点,利用匹配原则并借助于特征向量图形成匹配的特征点对,再以此为先验知识改进互信息相似性测度函数,最终通过自适应梯度下降优化方法寻找最优的配准参数,从而实现全心脏的自动分割。实验结果表明,论文提出的方法优于基于传统的互信息配准算法。

A nonrigid registration combining mutual information（MI）with marching cubes shape points is proposed.Those points are regarded as a prior knowledge of the shape landmarks of cardiac structure.Mutual information combined with feature point pairs is used as the similarity measure function,namely a cost function.All feature descriptors of 15 dimensions are used to find out matched point pairs between marching cubes points in atlas intensity images and another points in the neighborhood of unseen images needing segmentation.The adaptive stochastic gradient descent（ASGD）optimization is used to obtain optimal registration parameters.Two groups of experiments show that the proposed method achieves higher registration accuracy than traditional MI and more accuracy anatomical information of the whole cardiac structure can be gained.