基于微分同胚Demons形变配准算法获取肺通气图

摘要

基于四维计算机断层扫描图像(4D-CT)生成肺通气功能图的研究是医学图像处理领域的一个热点问题.肺通气图可以有效识别肺的不同功能区域,依此制定的肺癌个体化放疗计划可以有针对性的保护肺的高功能区,减轻放疗对肺功能的影响.本研究提出一种基于微分同胚Demons形变配准算法获取肺通气图的方法.首先使用阈值法对4D-CT呼气末时相和吸气末时相的肺体积进行分割;然后使用微分同胚Demons形变配准算法对两个时相分割出来的肺体积进行形变配准获取其形变函数Φ;最后应用雅可比行列式分析Φ,经重建获得肺通气图的三维分布.结果发现该方法充分考虑了图像的空间信息及灰度信息,无需借助其他功能影像设备,即可快速获取高分辨率的肺通气功能图像.%The study of the pulmonary ventilation image obtained on four-dimensional computed tomography (4D-CT) is a hot topic in medical image processing.The different functional regions of lung could be identified on pulmonary ventilation images.The individual radiotherapy plans considering differences in pulmonary functional regions can protect high functional region and reduce radiation-induced lung injuries.A novel approach based on diffeomorphic Demons registration algorithm was designed to obtain lung ventilation image.The lungs were segmented from the end-expiratory and end-inspiratory phase 4D-CT images using the threshold segmentation algorithm.The transform function Φ was obtained through the deformable registration of the lung volumes segmented from two phase CT images using diffeomorphic Demons registration algorithm.The function Φ was analyzed with the Jacobian model to obtain the three-dimensional distribution of lung ventilation images.The approach based on diffeomorphic Demons registration algorithm which considers the image spatial and gray information can be used to obtain high-resolution lung ventilation images quickly without extra function imaging equipments.