项目名称: 基于小框架的pMRI图像重建研究
项目编号: No.61303102
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 自动化技术、计算机技术
项目作者: 李炎然
作者单位: 深圳大学
项目金额: 27万元
中文摘要: 并行磁共振成像pMRI是一种对人体无损伤的医学成像技术,通过多个线圈采集部分目标数据加速成像速度,利用图像重建算法从获取的部分数据中恢复出目标图像。现有的pMRI机器仍然需要较长的成像时间,得到的图像可能会有伪影和噪声。图像的小框架系数在稀疏表示下能有效地表征图像边缘细节,滤除噪声。基于小框架的pMRI图像重建,减少成像时间,重建出高质量的MRI图像,是本项目拟研究的内容。 创新点包括:以新颖的研究视角将小框架理论应用于并行磁共振成像研究领域;研究近似l1凸优化模型与l0逼近算子闭型问题,为求解基于的非凸小框架系数稀疏优化模型提供新的理论方法与技术手段;研究pMRI图像重建的小框架正则化优化模型,探讨基于逼近算子的非线性不动点迭代算法,解决求解非凸不光滑的优化模型。
中文关键词: 并行磁共振成像;小框架;正则化;逼近算子;
英文摘要: Parallel magnetic resonance imaging (pMRI) is a non-invasive medical imaging technique that accelerates imaging speed through muliti-coils to sample parts of the target data and uses an image reconstruction algorithm to retrieve an image from the collected data. The imaging time of existing pMRI machine is still too long, and the reconstructed images may have aliasing and noise. Framelet coefficients of an image under sparse representation are effective to represent the image features and remove noise. The proposed research of framelet-based pMRI image reconstruction is to reduce MRI imaging time and reconstruct high-quality images. The main contributions of this proposal are as follows. 1) Framlet theory is applied to image reconstruction on pMRI with a novel research perspective; 2) The study of approximate convex model of l1 and the closed-form of the proximity operator of l0 provides new theory and technique for solving a non-convex framelet sparse optimization model; 3) Framelet-based regularization model of pMRI image reconstruction is studied and an iterative nonlinear fixed point algorithm using proximity operator theory is exploited to solve the non-convex and non-smooth optimization model.
英文关键词: pMRI;framelet;regularization;proximity operator;