PET-CT图像配准的预处理研究

Research on Preprocessing for PET-CT Image Registration

目的:目前,生物适形调强放疗(BIMRT)是放射治疗发展过程的一个重要转折点和突破。功能影像PET和解剖影像CT的图像配准在生物适形调强放疗中有非常重要的作用,本文旨在提出一种图像配准预处理方法并验证其可行性。方法:本文提出了一个PET和CT图像配准的预处理过程,该过程包括图像的归一化、基于模糊C均值聚类算法(FCM)和区域生长算法的图像分割、PET图像的分辨率调整以及图像的去噪与增强。结果:利用临床肺癌患者的PET和CT图像进行实验,结果表明该预处理过程能够很好地去除CT图像的扫描床等冗余信息以及PET图像的背景噪声,从而减少配准的复杂度,提高配准的速度和精度。结论:本文所提出的预处理方法是有效的,可以用来进行图像配准的预处理。

Objective： At present, biological intensity modulated radiation therapy is an important turning point and break- through for radiotherapy. The registration of fimctional image PET and anatomical image CT makes an important role in biological intensity modulated radiation therapy. An preprocessing method was proposed and verified in this paper. Methods： The preprocessing method proposed in this paper includes image normalization, image segmentation based on fuzzy c-means algorithm and regional growing algorithm, the resolution adjustment of PET image and the denoising of image. Results： The pro- posed method was verified by the processing of PET and CT image of lung cancer, and the result validated that some distur- bances such as scanning bed information of CT image and the noise of PET image can be removed efficiently. This method could reduce the complexity of the registration, and improve the registration speed and precision. Conclusions： The method proposed in this paper is efficient and can be used to perform image pre-processing.