基于离子群的快速图像配准与应用

Fast image registration based on particle swarm optimization algorithm and its application

图像配准在临床医学诊断和辅助手术过程中非常重要,采用仿X光和数字减影图像的数字重建影像(DRR)设计将使医学图像配准精度和速度提高,提出用片段着色器同时生成三维渲染图像和仿X光或者数字减影图像的DRR,将复杂优化和配准计算采用并行计算语言OpenCL完成,使整个DRR图像生成和2D/3D配准交由图像处理单元GPU完成。实验结果表明,依赖仿真的X光图像或者数字减影图像配准方法可为图像引导的手术提供精确和快速的辅助作用。

Image registration is critical for clinical medical diagnosis and assisting surgical procedure.Using simulated X-ray and digital subtraction angiography(DSA)to generate digitally reconstructed radiograph(DRR)would improve the accuracy and speed of medical image registration.Herein an approach of using shading language to simultaneously generate the 3D render image and DRR from simulated X-ray or DSA is presented.The complex optimization and registration calculations are completed using the parallel calculation language(OpenCL),and the image processing unit(GPU)is used for DRR generation and 2D/3D registration.Experimental results show that the proposed method which relies on simulated X-ray or DSA for image registration can provide rapid and accurate visual guiding for surgical procedure.