Combined motion-compensated and super-resolution image reconstruction to improve magnetic resonance imaging of the upper abdomen

项目目的展示使用呼吸运动建模作为提高临床腹部磁共振成像质量和成像效率的一种手段的好处。研究从减少的数据中生成运动替代信号以加速采集的方法，并研究使用这些信号对运动模型和运动校正图像的影响。调查减少“模型切片”的样本数量对运动模型和运动校正图像的影响。项目背景简介最近(由Jamie McClelland博士)演示了一个运动建模框架，该框架可以从部分图像数据(如2D图像切片)中建模3D运动。图像切片(模型切片)分别在呼吸周期内的多个点采集。该模型由一个或多个运动替代信号来参数化，所述一个或多个运动替代信号来自于在固定位置采集的“替代切片”。运动模型可用于运动补偿图像重建，以产生具有或不具有超分辨率(SR)重建的运动校正的3D图像。到目前为止，该框架仅应用于肺部，使用的是CT和MR成像。该项目将测试其在腹部MRI中的应用，特别是在肝脏成像中。腹部MRI通常涉及患者在获取每组图像时重复屏息。这限制了可获得的图像质量，使采集过程复杂化，降低了成像效率(由于屏气之间的恢复期)，并可能使患者感到困难或不舒服。本项目将采用临床用于肝脏结构成像的典型MR脉冲序列，并对其进行修改，以获取可用于估计肝脏运动场的信号。此字段将用于校正采集过程中发生的运动。超分辨率重建将在屏气获得的图像与自由呼吸和运动校正时获得的图像之间进行比较。

Aims of the projectDemonstrate the benefits of using respiratory motion modelling as a means to improve the image quality and imaging efficiency of clinical abdominal MRI.Investigate methods of generating motion surrogate signals from reduced data to speed up the acquisition and investigate the impact of using these signals on the motion models and motion corrected images.Investigate the effect of reducing the number of samples of the 'model slices' on the motion models and motion corrected images.Brief description of the project backgroundA motion modelling framework has recently been demonstrated (by Dr Jamie McClelland) that can model 3D motion from partial image data, such as 2D image slices. The image slices ('model slices') are each acquired at multiple points within the breathing cycle. The model is parameterised by one or more motion surrogate signals that are derived from a 'surrogate slice', which is acquired at a fixed location.The motion model can be used for motion compensated image reconstruction, to produce a motion corrected 3D image, with or without super-resolution (SR) reconstruction. The framework has, so far, only been applied in the lung, using CT and MR imaging. This project will test its application to abdominal MRI, in particular, to liver imaging.Abdominal MRI typically involves the patient holding their breath repeatedly as each set of images is acquired. This limits the image quality attainable, complicates the acquisition process, reduces the imaging efficiency (due to recovery periods between breath-holds) and can be difficult or uncomfortable for patients.This project will take a typical MR pulse sequence that is used clinically for structural imaging of the liver and modify it so that signals are acquired that can be used to estimate the motion field across the liver. This field will be used to correct for the motion that occurs during the acquisition. Super-resolution reconstructions will be compared between the images acquired with breath-holds and those acquired during free-breathing and motion-corrected.