基于Research Workplace的大鼠脑部PET/CT-MRI图像配准技术

PET/CT-MRI image registration in rat brain based on Research Workplace

目的随着脑计划的发展,动物脑部结构图像与脑部功能图像融合分析已经成为重要的研究课题。目前动物脑部影像主要有X线计算机断层扫描(computed tomography,CT)图像、正电子发射断层扫描(positron emission tomography,PET)图像、磁共振成像(magnetic resonance imaging,MRI)。Micro-CT/PET图像中CT图像不能清楚地显示每个脑区结构,不能准确测得PET功能图像的数据,为了解决这一缺点,提出将动物脑部MRI图像与PET图像配准的技术。方法利用Bruker7.0T核磁采集到的大鼠脑部T2WI像制作一套MRI脑部结构模板数据,把大鼠MRI脑部结构模板数据导入到Research Workplace软件中,通过该软件提供的数据处理模块对MRI大鼠脑部模板与10套大鼠PET/CT数据做仿射拉伸、旋转,使图像在同一空间内。通过封装的最大互信算法以及Powell优化算法对这些图像进行配准。结果将PET/CT配准图像作为配准标准,默认配准精度是最高的。引入图像统计特性的评价参数证明MRI/PET配准图像精度和原始PET/CT图像是相近的。Research Workplace软件在大鼠脑部MRI图像与PET/CT图像配准精度上可达到要求。结论基于Research Workplace的大鼠脑部PET/CT-MRI图像配准技术可以解决PET/CT难以准确获取功能像数据的缺点,并能达到配准精度要求。

Objective The fusion analysis of animal brain structural images and functional images has become an important research area with the development of the China brain plan. At present,animal brain images mainly contain computed tomography(CT) images,positron emission tomography(PET) images and magnetic resonance imaging(MRI). Since the images of Micro-CT/PET cannot clearly show the structure of each brain region,the data of PET functional image cannot be accurately measured. In order to solve this shortcoming,it is proposed to use the animal brain MRI image and PET image registration to accurately measure the functional PET data. Methods A set of rat brain structural MRI template data was generated by using the T2WI image of rat brain collected by Bruker 7.0T nuclear magnetic resonance. Then,the rat brain structural MRI template data was imported into Research Workplace software and processed by internal module of Research Workplace. The rat brain MRI template was affine stretched and rotated with 10 sets of rat PET/CT data to make the image in the same coordinate. Then,these images were registered by the encapsulated maximum mutual trust algorithm and the Powell optimization algorithm. Results The PET/CT registration image was defined as a registration standard with the highest registration accuracy. The evaluation parameters of the statistical properties of the image were introduced to demonstrate that the MRI/PET registration image accuracy was similar to the original PET/CT image. The Research Workplace software fully met the requirements for the registration accuracy of MRI images and PET/CT images in rat brain. Conclusions PET/CT-MRI image registration in rat brain based on Research Workplace can solve the shortcomings of PET/CT, which is difficult to accurately acquire functional image data, and can meet the requirements for registration accuracy.