基于QPSO算法的3D多模医学图像配准

3D multi-modality medical image registration based on QPSO algorithm

基于互信息的配准方法具有精度高、鲁棒性强的特点。但基于互信息的目标函数存在许多局部极值,给配准的优化过程带来了很大的困难。把量子行为的粒子群优化算法(QPSO)应用到了3D医学图像配准中。QPSO不仅参数个数少,其每一个迭代步的取样空间能覆盖整个解空间,因此能保证算法的全局收敛。实验结果表明,该算法能够有效地克服互信息函数的局部极值,大大提高了配准精度,与美国Vanderbilt大学的"金标准"比较,达到了亚像素级的精度。

Image registration based on mutual information is of high accuracy and robustness.Unfortunately,the mutual information function is generally not a smooth function but one containing many local maxima,which has a large influence on optimization.This paper proposes a registration method based on Quantum-behaved Particle Swarm Optimization（QPSO） algorithm.Not only QPSO has less parameters to control,but also does its sampling space at each iteration covers the whole solution space.Thus QPSO can find the best solution quickly and guarantee to be global convergent.Experiments show that this registration method can efficiently restrain local maxima of mutual information function and improve accuracy.Compared with the gold standard,the sub-pels accuracy can be achieved.