A new computational tool for the analysis of Floor Borne Vibrations on the performance and image quality of MRI scanners

MRI扫描仪的功能依赖于超导磁体（即主线圈）产生的高强度和极其均匀的磁场，以及交流梯度线圈中指定的时变电流特征产生的非均匀磁场。核磁共振成像扫描仪的第三个关键部件是低温恒温器，它由一系列辐射屏蔽器组成，并使主磁铁线圈浸入氦气容器内的液氦中。相反，在室温下，梯度线圈位于低温恒温器外。不幸的是，外部产生的振动，也被称为地板振动（FBV），在磁铁上引入了不希望的加速度（主要是在垂直方向上，但也有一些在横向方向上）。这些振动会导致辐射屏蔽层和磁体之间的相对运动，从而在屏蔽层中产生不必要的涡流，进而产生二次磁场，后者影响初级均匀磁场的质量，最终导致不理想的成像伪影。挑战：为了减轻FBV的负面影响，在低温恒温器下面建造了一定程度的隔振磁铁，以软橡胶垫或弹簧阻尼器组件的形式，或者通过反复试验改变磁铁和低温恒温器界面的机械行为。然而，在现实的3D MRI配置中，这些隔振装置的优化设计和精确定位是一项极其复杂的任务，需要专家的人为干预。当前的挑战包括预测MRI扫描仪在大范围输入加速度频率下的性能，并通过(a)材料参数的最佳选择（即屏蔽电导率，碳纤维悬架刚度）和(b) MRI扫描仪组件的形状优化（即辐射屏蔽组件的平端或圆端）来确定降低其对FBV灵敏度所需的工程变化。这需要对FBV对：(a)磁场分布（即十亿分之一的水平）和(b)输出图像质量的影响有先验的（并且非常准确的）了解。这只能通过尖端的高保真的计算机建模工具来实现，对可用的实验数据进行稳健的基准测试，并通过使用降阶建模（ROM）技术嵌入西门子Healthineers的设计周期，该技术可以允许材料参数和/或几何特征的快速变化。目的：开发一种新的鲁棒，准确和快速的数据驱动的3D硅ROM计算框架，用于FBV对MRI性能和成像质量的建模。这个具有挑战性的EPSRC案例奖项目提案的目标有四个：1。当受到外部FBV时，在现实的3D MRI配置中精确计算涡流和磁场（即十亿分之一的水平）。这里将利用高阶有限元的使用，研究产生所需精度水平所需的内插的“p”阶。2. 对软件工具进行基准测试，对比从埃尔兰根和牛津大学最先进的振动台收集的实验数据。实验数据有助于保证软件的鲁棒性和可靠性。通过将开发的软件工具与Siemens healthineers的内部成像分析工具相结合，研究FBV对图像质量的影响。在考虑材料参数（即刚度，电导率）和/或几何特征（即屏蔽几何）的快速变化时，开发用于快速多次查询的ROM技术。西门子Healthineers的内部经验将在这里被利用，以促进对FBV最敏感参数的识别。

THE CONTEXTThe functioning of an MRI scanner relies on high strength and extremely uniform magnetic fields generated through superconducting magnets (i.e. main coils) and non-uniform magnetic fields generated by time-varying current signatures specified in AC gradient coils. A third key component in an MRI scanner is the cryostat, which is comprised of a series of radiation shields and keeps the main magnet coils immersed in liquid Helium within a Helium vessel. On the contrary, the gradient coils sit outside the cryostat at room temperature. Unfortunately, externally generated vibrations, also known as Floor Borne Vibrations (FBV), introduce undesirable accelerations on the magnets (primarily in the vertical direction, but also some in lateral direction). These vibrations can lead to relative movement between the radiation shields and the magnets, thus generating unwanted eddy currents in the shields which, in turn, produce secondary magnetic fields, the latter affecting the quality of the primary uniform magnetic field and, ultimately, resulting in non-desirable imaging artefacts.THE CHALLENGETo alleviate the negative impact of FBV, magnets are built with some amount of vibration isolation either underneath the cryostat, in the form of either a soft rubber matting or a spring-damper assembly, or by trial and error changes to the mechanical behaviour of the magnet and cryostat interface. However, the optimal design and the precise location of these vibration isolation devices within a realistic 3D MRI configuration is an extremely complex task which requires expert human intervention. The challenge at hand consists of predicting the performance of the MRI scanner over a wide range of input acceleration frequencies and determining the engineering changes required to reduce its sensitivity to FBV, via (a) the optimal selection of material parameters (i.e. shield conductivity, carbon fibre suspension stiffness) and (b) the shape optimisation of the MRI scanner components (i.e. flat or round ends in radiation shield components). This requires the a-priori (and very accurate) knowledge of the effect of FBV on: (a) the magnetic field distribution (i.e. to the Parts Per Billion level) and (b) the output image quality. This can only be achieved via cutting-edge high-fidelity in-silico modelling tools, robustly benchmarked against available experimental data, and embedded within the design cycle at Siemens Healthineers via the use of Reduced Order Modelling (ROM) techniques which can permit the rapid variation of material parameters and/or geometrical features.THE AIMThe development of a new robust, accurate and fast data-driven 3D in-silico ROM computational framework for the modelling of FBV on MRI performance and imaging quality. THE OBJECTIVESThe objectives of this challenging EPSRC CASE Award project proposal are four:1. The accurate computation of eddy currents and magnetic fields (i.e. to the Parts Per Billion level) within a realistic 3D MRI configuration when subjected to external FBV. The use of high order Finite Elements will be here exploited, investigating the order "p" of interpolation needed to produce the level of accuracy required. 2. The benchmarking of the software tool against experimental data collected from state-of-the-art shaker tables at Erlangen and Oxford. Experimental data will help ensure robustness and reliability of the software.3. The study of the impact of FBV on image quality through the combination of the developed software tool with in-house imaging analysis tools at Siemens Healthineers.4.The development of a ROM technique for fast multiple-query when considering the rapid variation of material parameters (i.e. stiffness, conductivity) and/or geometrical features (i.e. shield geometry). In-house experience at Siemens Healthineers will be here exploited in order to facilitate the identification of the most sensitive parameters to FBV.