项目名称: 多尺度分析技术在图像引导肿瘤精确放疗中的研究及临床应用
项目编号: No.61201441
项目类型: 青年科学基金项目
立项/批准年度: 2013
项目学科: 电子学与信息系统
项目作者: 李登旺
作者单位: 山东师范大学
项目金额: 25万元
中文摘要: 本项目的研究目的是结合多尺度理论来研究医学图像形变配准和纹理分析新技术,并在肿瘤精确放疗中进行临床应用。形变配准技术是实现图像引导自适应放疗的关键,针对采用互信息作为相似性测度的形变配准算法计算量较大,且缺少空间位置信息等缺陷,本项目提出具有边缘保护性质的多尺度配准框架,以此实现计划CT与日常放疗CBCT图像配准,在保证精度的前提下达到减少计算量的目的,从而有效指导放疗摆位和自适应放疗。同时,为跟踪放疗过程中肿瘤的变化,本项目选取具有多尺度和多方向性质的双树复小波变换来分别表达放疗过程中4D-CBCT图像的全局和局部形变,并使用Naiver偏微分方程设计极小化能量函数实现小波系数的估计。另外,为帮助临床医生制定更为合理的放疗计划,采用多尺度技术提取PET图像中肿瘤内纹理特征,并量化肿瘤异质性程度。本项目所开展的研究,对于医学图像处理算法理论研究和临床肿瘤精确放疗应用均具有重要意义。
中文关键词: 多尺度分析;形变配准;肿瘤纹理分析;图像引导放疗;
英文摘要: The purpose of the project is to embed the multi-scale theory into the deformable registration and texture analysis algorithms in distinct medical images, and investigate the clinical applications in tumor precision radiotherapy and image guided radiotherapy. Currently deformable registration is an essential technique to perform the image guided adaptive radiotherapy. However, similarity measure based on the utilization of mutual information results in the tremendous increasing computational time and disregards the importance of spatial information. We propose the novel multi-scale framework based on edge preserving scale space to achieve the accurate and efficient registration of planning CT and daily CBCT. The proposed framework can improve registration speed without depressing the accuracy, and secure the effective guidance for radiotherapy positioning and adaptive radiotherapy. Furthermore, to track changes of tumors in the radiotherapy process, dual tree complex wavelet transform which has the advantage of multi-scale and multi-directional properties, is used to express the global and local deformation of the 4D-CBCT images during radiotherapy. In our framework, the energy function is formulated to estimate the wavelet coefficients by using Naiver partial differential equations. In addition, to assist clini
英文关键词: multiscale analysis;deformable registration;tumor texture analysis;image guided radiation therapy;