基于傅里叶梅林变换的图像配准方法

Fourier-Mellin transformation-based method for image registration

目的:提出一种新的配准框架用于图像引导放射治疗系统中的2D/3D图像配准,有效降低传统方法迭代搜索时间,同时保证放射治疗要求的配准精度。方法:利用傅里叶梅林变换方法对正侧位kV图像与对应方位参考CT图像生成的数字重建放射影像(DRR)进行粗配准,根据傅里叶梅林变换计算得到的二维平移向量以及放射治疗系统的机械几何参数反推出参考CT图像的三维空间位置偏差,更新正侧位的DRR图像,最后通过正侧位kV图像与DRR图像的相似度进行精配准达到临床需求。结果:采用临床金标准数据验证方法的配准性能,实验结果表明,配准误差为0.5765 mm,平均运行时间为3.34 s。结论:该方法鲁棒性强,对图像的噪声不敏感,人工干预少,可满足临床应用的需求。

Objective To propose a novel 2D/3D image registration frame in image guided radiotherapy system for reducing the time for iterative search and ensuring the registration accuracy required for radiotherapy.Methods Fourier-Mellin transformation algorithm was adopted for coarse registration between kV digital radiographs and the corresponding digitally reconstructed radiographs(DRR)generated by CT images in posteroanterior view and lateral view.The 3D position errors of CT images which were estimated based on the 2D translation vector calculated by Fourier-Mellin transformation and the geometric parameters of the radiotherapy system were used for updating the DRR in posteroanterior view and lateral view,and the precise registration between kV images and the DRR images was completed using image similarity metric.Results The validation on the clinical gold standard data set showed that the registration error was 0.5765 mm and that the average running time was 3.34 s.Conclusion The proposed method is suitable for clinical application with good robustness,insensitive to image noise and less human intervention.