基于多层次变换和优化方法的PET-CT图像弹性配准

PET-CT image elastic registration based on multi-transformation and multi-optimizer

目的为校正PET/CT一体化仪器的误配准和形变问题。方法本文提出一种基于多层次变换和优化的弹性配准方法,以互信息为相似性测度,先进行全局刚性变换使用单纯形优化策略,在刚性变换的基础上使用B样条形变变换并使用有限内存的BFGSB(limited memory Broyden Fietcher Goldfarb Shanno bound,LBFGSB)优化算法。选取3个患者PET/CT联合扫描下的6组数据对本文算法进行验证,并与采用传统单一变换方法和单一优化策略的方法进行了对比。结果采用本文提出的配准算法得到的平均配准时间为19.78 s,平均互信息值为0.3115,较传统仅使用单一变换和单一优化的配准方法在配准速度和精度上有一定的提高。结论基于多层次变换和优化方法的图像配准方法能够较好地解决PET产生的形变问题,配准速度快精度高,可以用来校正PET/CT联合扫描由于呼吸运动和脏器运动产生的误配准和形变问题。

Objective To correct the registration error and deformation of the combined PET/CT scanner. Methods This paper proposed an elastic registration method based on multi-transformation and multi- optimizer by using mutual information as metric. This method adopted rigid transformation firstly to ensure the global registration with simplex optimizer and then used BSpline FFD with limited memory Broyden Fietcher Goldfarb Shanno bound（LBFGSB） optimizer. This novel method was compared with those methods based on single transformation and single optimizer through testing six image pairs of three patients under combined PET/ CT scanner. Results The mean registration time of the proposed method was 19.78 s and the mean value of mutual information was O. 3115,which was better than those methods based on single transformation and single optimizer. Conclusions The proposed method based on multi-transformation and multi-optimizer could handle the deformation and the registration error caused by respiratory and organ motion in the combined scannerimages and was superior to other traditional methods in speed and accuracy.