基于流体映射模型的医学图像弹性配准

Medical Image Elastic Registration Based on Fluid Diffeomorphisms

为了实现约束条件下的非刚性匹配,引入流体映射模型,以两组反映待配准图像轮廓信息的对应控制点集,对头颅,肝,脾等各种器官进行弹形匹配。流体模型借用了流体运动中流体的相互关系的数学模型,通过寻求控制点集的最优变换轨迹,以实现图像各象素点相互制约的平滑变形。实验结果证明,无论是对小变形还是大变形图像的弹性配准,使用流体映射模型都能取得良好的配准效果。

Fluid diffeomorphisms model is introduced to realize the no-rigid registration based on two control sets of corresponding points for head, liver, spleen, etc. The smooth deformation of constrained image point is got by seeking optimal lagrangian trajectory of control sets, using the mathematic model borrowed from fluid physics, which defines the connection of moving fluid. Experiment results showed that good registration was obtained for both small and large deformation.