非刚体图像配准的变形场拓扑约束研究

利用正则化方法约束非线性变形场是非刚体图像配准领域的一个重要研究方向。为得到具有拓扑保持能力的非线性变形场，本文在分析粘流体配准和扩散模型配准算法的实现原理基础上，提出一种基于弹簧约束的变形场拓扑保持方法。该方法在可变形图像上附加不规则网格，通过保持网格结点间的连接关系不变达到控制图像变形的目的。将本文算法应用在不同人脑磁共振图像配准和脑内核结构分割中，结果表明，本文方法具有保持变形场拓扑不变的能力，且能够给出更为准确的分割结果。 

Research of Deformation Field Topology Constraint Method in Non-rigid Image Registration

Regularizing nonlinear deformation field is an important aspect in non-rigid medical image registration. By analyzing the implementation principles of vicious fluid registration and diffusion registration method, this paper proposed to preserve the topologies of the deformation field using a spring constraints. The template image is covered with a triangular mesh, and the connection relationship between mesh vertices is preserved. The method is evaluated by registering cerebral MRI (magnetic resonance imaging) image data obtained from different individuals and by segmenting internal brain structures. Experimental results show that the proposed method has good deformation ability and topology preserving ability. The segmentation results are more accurate.