Marami Bahram, Sirouspour Shahin, Capson David W
Department of Electrical and Computer Engineering, McMaster University, 1280 Main Street West, Hamilton, ON, Canada, L8S-4K1.
Med Image Comput Comput Assist Interv. 2011;14(Pt 1):460-7. doi: 10.1007/978-3-642-23623-5_58.
We have developed an automatic model-based deformable registration method applicable to MR soft-tissue imaging. The registration algorithm uses a dynamic finite element (FE) continuum mechanics model of the tissue deformation to register its 3D preoperative images with intraoperative 1) 3D low-resolution or 2) 2D MR images. The registration is achieved through a filtering process that combines information from the deformation model and observation errors based on correlation ratio, mutual information or sum of square differences between images. Experimental results with a breast phantom show that the proposed method converges in few iterations in the presence of very large deformations, similar to those typically observed in breast biopsy applications.
我们开发了一种基于模型的自动可变形配准方法,适用于磁共振软组织成像。该配准算法使用组织变形的动态有限元(FE)连续介质力学模型,将术前3D图像与术中1)3D低分辨率图像或2)2D磁共振图像进行配准。配准通过一个滤波过程实现,该过程结合了来自变形模型的信息和基于图像间相关比、互信息或平方差之和的观测误差。使用乳腺模型的实验结果表明,在存在非常大变形的情况下,该方法在几次迭代中即可收敛,这种大变形类似于在乳腺活检应用中通常观察到的情况。