肝脏4D动态对比增强磁共振成像的图像重建和配准
发布时间:2018-04-20 20:29
本文选题:肝脏4D + DCE-MRI ; 参考:《浙江大学》2017年硕士论文
【摘要】:肝细胞癌是异质性最强、死亡率最高的恶性肿瘤之一。在肝硬化结节逐步形成肝细胞癌的过程中,新生异常动脉血供的增多是临床诊断的一个重要判据。肝脏四维动态对比增强磁共振成像(DCE-MRI:Dynamic Contrast-Enhanced Magnetic Resonance Imaging)不仅可提供病变组织形态学的信息,还可反映病变组织血流微循环的改变,从而为肝细胞癌的早期诊断及疗效评估提供更加丰富、精准的影像信息。肝脏4D DCE-MRI的临床应用主要面临两大技术挑战:一是快速成像,需要基于高度降采样的k空间数据,重建高时空分辨率的全肝4D图像;二是呼吸运动校正,需要有效去除长时间扫描中呼吸运动引起的伪影。本文重点研究自由呼吸下的高时空分辨率全肝4D DCE-MRI的关键技术,主要取得了如下创新成果:(1)针对基于高度降采样的A空间数据重建高时空分辨率图像的需求,提出了一种基于字典学习的低秩稀疏分解重建算法。采用低秩稀疏分解模型有效挖掘重建图像帧之间的相关性,利用字典学习进一步增强算法对数据的自适应性,从而实现了图像信号的高稀疏性表达。实验表明,本文算法可以有效去除降采样伪影,改善重建图像细节。(2)面向自由呼吸扫描,提出了一种内嵌呼吸运动校正的动态MRI重建算法。首先,基于A空间数据直接估计一维呼吸运动信号,并据此信号k将空间数据划分到若干运动状态中;然后,基于各状态的A空间数据,采用低秩稀疏分解分别重建图像序列;最后,采用稀疏性约束对整个序列做二次重建。实验表明,本文算法可以有效去除运动伪影,重建自由呼吸下的高时空分辨率图像。(3)针对胸腹部图像的呼吸运动配准问题,提出了 一种保持运动场不连续性的配准算法。采用基于马尔可夫随机场的离散优化方法,结合运动场估计值和图像信号自动分割得到内脏与胸腹部内壁之间呼吸滑动的界面,然后去除横跨界面的运动场平滑性约束。实验表明,本文算法可以实现呼吸滑动界面的自动分割,有效降低配准误差。
[Abstract]:Hepatocellular carcinoma (HCC) is one of the most heterogeneous malignant tumors with the highest mortality. The increase of abnormal arterial blood supply is an important criterion for clinical diagnosis of hepatocellular carcinoma (HCC). DCE-MRI: dynamic Contrast-Enhanced Magnetic Resonance imagingcan not only provide information of pathological morphology, but also reflect the changes of blood flow microcirculation in pathological tissues, thus providing more valuable information for early diagnosis and evaluation of curative effect of hepatocellular carcinoma. Accurate image information. The clinical application of liver 4D DCE-MRI faces two major technical challenges: one is rapid imaging, which requires reconstruction of the whole liver 4D image with high spatiotemporal resolution based on highly decimated k spatial data, and the other is respiratory motion correction. It is necessary to effectively remove artifacts caused by respiratory movement during long-term scanning. This paper focuses on the key technologies of high spatiotemporal resolution 4D DCE-MRI under free breathing. The main achievements are as follows: 1) the requirement of reconstruction of high spatiotemporal resolution images based on A spatial data based on highly decimated sampling. A low rank sparse decomposition reconstruction algorithm based on dictionary learning is proposed. The low rank sparse decomposition model is used to effectively mine the correlation between reconstructed image frames, and the dictionary learning is used to further enhance the adaptability of the algorithm to the data, thus realizing the high sparsity representation of image signals. Experimental results show that the proposed algorithm can effectively remove the subsampling artifacts and improve the image details for free breathing scanning. A dynamic MRI reconstruction algorithm with embedded breathing motion correction is proposed. Firstly, the one-dimensional breathing motion signal is directly estimated based on A spatial data, and then the spatial data is divided into several moving states based on the signal k, and then the image sequence is reconstructed by low-rank sparse decomposition based on the A-space data of each state. Finally, the sparse constraint is used to reconstruct the whole sequence twice. Experimental results show that the proposed algorithm can effectively remove motion artifacts and reconstruct high spatial and temporal resolution images under free breathing. To solve the problem of respiratory motion registration in chest and abdomen images, a registration algorithm to maintain discontinuity of sports fields is proposed. A discrete optimization method based on Markov random field is used to segment automatically the respiratory sliding interface between the viscera and the inner wall of the chest and abdomen by combining the estimation value of the sports field and the image signal. Then the smoothness constraint of the motion field across the interface is removed. Experiments show that the proposed algorithm can realize automatic segmentation of respiratory sliding interface and reduce registration error effectively.
【学位授予单位】:浙江大学
【学位级别】:硕士
【学位授予年份】:2017
【分类号】:R735.7;TP391.41
【相似文献】
相关期刊论文 前2条
1 熊海涛,胡匡祜,苏万芳,李淑宇,苏德森,顾学裘;脂质体、乳剂图像自动定量分析方法[J];药学学报;2000年08期
2 韩莉,
本文编号:1779303
本文链接:https://www.wllwen.com/yixuelunwen/zlx/1779303.html