用改进的粒子群算法实现多模态刚性医学图像的配准

Multimodality Rigid Medical Image Registration Through Improved Particle Swarm Optimization

多模态医学图像的配准在医学诊断和治疗计划中起着重要的作用 提出了一种基于轮廓特征点和改进的粒子群优化算法 ((ParticleSwarmOptimization ,PSO) )求解的配准方法 该方法首先用Canny算子提取图像的边缘 ,用ISODATA算法进行聚类分析提取出轮廓特征点 ,然后用两轮廓匹配点对的欧几里德距离平均值的极小值作为两个特征点对配准准则 ,并用改进的PSO算法求解配准所需的空间变换参数 实验证明 ;该方法配准精度能够达到亚像素级 ,能够避免陷入局部极小值而且速度得到明显改善 。

Multimodality medical images registration based on contour feature points and improved particle swarm optimization is presented for clinical diagnosis and therapy planning. The image edges are first detected by using Canny operator, then the contour feature points are extracted by ISODATA algorithm, and translation parameters are calculated by using improved particle swarm optimization algorithm. Experiments show that this approach is efficient and can avoid local minimum.