MR image reconstruction for non-Cartesian data and quantitative susceptibility maps using generalized sampling

本项目旨在改进非笛卡尔核磁共振数据的数据采集和重建，特别是在定量敏感性成像的背景下。越来越多的磁共振数据采用非笛卡儿方式获取，与经典笛卡儿数据采集相比，非笛卡儿数据采集具有诸多优势，定量易感成像这一新兴领域在神经放射学领域得到越来越多的应用，例如脑肿瘤、髓鞘和出血的成像。虽然已经报告了有希望的初步结果，但当深入观察时，存在许多困难。例如，笛卡尔序列容易受到运动伪影的影响，需要较长的采集时间，重建算法容易受到伪影的影响。在这个项目中，应该开发健壮的数据采集和重建，这些数据采集和重建没有人工制品。本项目包括数据采集和重建两个方面。在采集序列中对广义采样重构的最优采样方案进行了仿真和实现。利用小波和剪切波，设计基于广义采样的重建方案。这些方案也将用于定量敏感性图的重建。此外，非傅立叶编码的数据应该用射频脉冲再现以前使用的小波和shearlet。这些非傅立叶编码的数据将依次由先前开发的重构算法进行重构。

This projects aims to improve data acquisition and reconstruction for non-Cartesian MR-data, especially in the context of quantitative susceptibility imaging. An increasing amount of MR data is acquired non-Cartesianly which has various advantages compared to a classical Cartesian data acquisition, and the rather new field of quantitative susceptibility imaging is getting applied more and more in the area of neuroradiology, e.g. in imaging brain tumors, myelin and hemorrhage. Although promising first results have been reported, when viewed in depth, there are many difficulties. For example, the Cartesian sequences are susceptible to motion artifacts and require long acquisition times, and reconstruction algorithms are prone to artifacts. In this project, robust data acquisition and reconstructions which are free of artifacts shall be developed. This project includes aspects of data acquisition and reconstruction. Optimal sampling schemes for the generalized sampling reconstruction are to be simulated and implemented in acquisition sequences. Using wavelets and shearlets, reconstruction schemes based on generalized sampling will be designed. These schemes will also be used for the reconstruction of quantitative susceptibility maps. Furthermore, non-Fourier encoded data should be acquired with RF-pulses reproducing the previously used wavelets and shearlets. This non-Fourier encoded data shall, in turn, be reconstructed by the previously developed reconstruction algorithms.