面向虚拟手术建模的三维磁共振弹性成像方法研究

Robot-assisted surgery (RAS) is one of the fastest evolving techniques in clinic. Owing to its competency of minimal invasion and microscopic operation, RAS is now transforming the practices of clinical surgery greatly. Most surgeons have to upgrade their knowledge and skills in time. Surgical training robots come into being and provide an augmented scenario for convenient training and practicing. It is now possible to create various digital organs from clinical scans of computed tomography (CT) or magnetic resonance imaging (MRI). However, neither CT nor MRI is able to provide the biomechanical information of organs and tissues. In other words, it yet suffers from the deficient information of haptics in establishing surgical training robots...There have been a few solutions, including ultrasound elastography and magnetic resonance elastography (MRE), for invasive acquisition of biomechanical information. MRE is promising for quantitative measurement and arbitrary imaging window. With the generous support of the National Natural Science Foundation of China (NSFC), we have successfully established a prototypical MRE system that is applicable to regional evaluation of organs and tissues. However, such localized measurement is not efficient for modeling augmented human body. It thus calls for a fully three-dimensional (3D) MRE...Many challenging issues remain for a 3D MRE system, for example, inhomogeneous coverage of tissue actuation, low efficiency of wave imaging and insufficient robustness of elasticity reconstruction. In this study, we first propose to develop a cohort of actuators, each of whom has individual vibrating frequency and phase. The optimal organization of these actuators might lead to homogeneous actuation of multiscale organs and tissues. In the second, a new mechanism is proposed to accelerate MRE imaging in parallel. It will replace the balanced bipolar gradient pairs with an echo chain of reading gradient pairs. The latter breaks the limitation of short spin-spin relaxation. This new imaging pulse sequence will be further enhanced by integrating fractional encoding. It is advantageous for rapid acquisition of multifrequency, multislice and multidirectional wave images. Finally, a new algorithm is proposed to process 3D wave images as a whole and reconstruct a hybrid shear modulus distribution. It takes the Helmholtz-Hodge method to decompose 3D wave fields. Then, the low-rank algebraic is applicable for noise-robust elasticity reconstruction...This study is concentrated on the new methods for 3D MRE. The outcomes are contributive to establishing a prototypical system for noninvasive acquisition of 3D elastic information of organs and tissues. It is of vital importance for modeling haptic-augmented virtual surgery.

本项目面向虚拟手术建模与仿真训练之应用需求，针对数字器官缺乏有效触感这一核心问题，致力于研究三维磁共振弹性成像方法与技术，解决其驱动、成像与解析方面的一些关键科学问题，具体包括：①针对弹性波探针空间分布不均所造成的低效驱动问题，将研究多源多相组合驱动机制，从理论上指导三维稀疏驱动阵列的设置与优化；②针对反复采样所造成的成像耗时过长问题，将研究分形编码多级梯度回波并行成像机制，发展高效快速磁共振弹性成像扫描序列；③针对算法不完备所造成的非鲁棒性重构问题，将以亥姆霍茨—霍奇方法为核心，建立低阶波动方程求逆和三维一体混合弹性重构算法。..本项目以前一基金项目成果—吸附载波驱动机制—为基础，创新磁共振弹性成像技术，研发三维弹性成像系统，从而无创获取多尺度组织器官的力学属性，建立组织结构—力学属性耦合模型，渲染出组织结构清晰、触感层次分明的虚拟手术环境，推动机器人辅助手术训练技术的发展。

本项目面向虚拟手术建模与仿真训练之应用需求，针对数字器官缺乏有效触感这一核心问题，研究了三维磁共振弹性成像的方法与技术，尤其是驱动、成像与解析等方面，具体包括：①瞬态型核磁共振弹性成像方法仿真研究，并且搭建了开源型 MREJ 核磁共振弹性成像实验平台，在永磁型低场核磁共振弹性成像实验平台上进行了仿体和人体实验；②提出了一种具有负磁导率的磁感透镜，通过使用超材料来提高MRI表面线圈的信噪比；③基于主成分分析和深度学习建立了新型的医学图像辅助诊断技术。