MAGNETIC RESONANCE IMAGING OF MOVING PATIENTS AT ULTRA-HIGH FIELD: REAL-TIME MOTION CORRECTED PARALLEL-TRANSMIT PULSE DESIGN

超高场(UHF、7T及以上)磁共振成像(MRI)扫描仪提供了一个独特的机会，可以比以前更高的分辨率研究大脑。许多脑部核磁共振成像采集持续时间长达几分钟。在此期间，由于不自主的运动或呼吸，对象位置的任何微小偏差都可能改变成像的体积，导致图像中的破坏性伪影，并需要重新扫描患者。此外，由于运动对MRI过程中发生的少量组织加热的影响还没有很好地了解，因此出现了关于患者安全的问题。运动对于不合作的患者尤其有问题，比如在儿科成像中，或者对于帕金森氏症或痴呆症患者。镇静是这种情况下的常见做法，它是侵入性的，会影响非结构性磁共振成像技术(如功能磁共振成像)所获得的结果，据报道会导致不良副作用，甚至进入急救。为了克服UHF的电磁干扰，可以使用并行传输(PTX)硬件，显著改善图像质量和对组织加热的控制。PTX允许射频(RF)脉冲--用于产生MRI信号--通过多个独立控制的RF通道传输。然而，PTX的好处是以加剧上述与运动相关的问题为代价的，因为通道的RF以复杂的方式相互作用。PTX提供的额外自由度(即通道是独立控制的)意味着，如果更全面地了解运动条件下通道之间的相互作用，并将其应用于RF脉冲设计过程，则可以通过一种新的PTX RF脉冲设计方法来解决这个问题。首先，我正在开发一种运动稳健的方法来设计PTX脉冲。利用计算机模拟，我最初调查了不同PTX环境下运动的影响，以更好地了解所涉及的动力学。在此之后，我设计了一种新的方法，在模拟中降低了PTX脉冲对运动的敏感性，这意味着即使在患者运动的情况下，图像质量仍然很高。我的下一步是继续模拟，以确保安全(组织加热)问题也从我的方法中受益，并使用扫描仪实验验证该方法。除了通过引入运动稳健性来降低灵敏度之外，我研究的第二个重点是实时PTX脉冲设计方法。这将允许将实时更新应用于扫描仪，在整个扫描过程中补偿患者的运动，因此更准确和有效地消除所有运动诱导的影响。我正在使用一种机器学习方法来训练神经网络，用模拟的训练数据来开发这种方法，然后在实验中进行测试。我的工作的影响是广泛的。通过消除与PTX运动相关的担忧，UHF MRI的好处(例如，更高的图像分辨率)可以应用于临床。癫痫和多发性硬化症的诊断和预后等领域已经在研究环境中证明了这一点的好处，但这里提到的技术挑战目前阻碍了它的广泛或临床应用

Ultra-high field (UHF, 7T and above) magnetic resonance imaging (MRI) scanners present a unique opportunity to study the brain at much higher resolution than previously possible. Many brain MRI acquisitions last upwards of several minutes. Any small deviation in subject position during this time, due to involuntary movement or breathing, can change the imaged volume, leading to destructive artefacts in the images, and necessitating a rescan of the patient. Additionally, questions arise regarding patient safety, as effects of motion on the small amounts of tissue heating which occur during MRI are not well understood. Motion is specifically problematic with uncooperative patients such as in paediatric imaging, or for patients with Parkinson's or dementia. Sedation, which is common practice in such cases, is invasive, affects results obtained for non-structural MRI techniques (e.g. functional MRI), and has been reported to cause adverse side effects and even admittance to emergency care. To overcome electromagnetic interference at UHF, parallel transmission (pTx) hardware can be used, significantly improving image quality and control over tissue heating. pTx allows radiofrequency (RF) pulses -used to generate the MRI signal- to be delivered by multiple, independently controlled RF channels. However, the benefits of pTx come at a cost of exacerbating the motion-related problems described above, as the channels' RF interacts in complex ways. The extra degrees of freedom offered by pTx (i.e. because the channels are independently controlled) mean that the problem could be overcome with a new approach to pTx RF pulse design, given a more comprehensive understanding of the interactions between channels under conditions of motion, and application of that to the RF pulse design process.My research approaches this problem in two ways. Firstly, I am developing a motion-robust approach to pTx pulse design. Using computer simulations, I initially investigated the effects of motion in different pTx contexts to better understand the dynamics involved. Following this, I devised a new approach which has reduced the sensitivity of pTx pulses to motion in simulations, meaning that image quality remains high even in the case of patient motion. My immediate next steps are to continue simulations to ensure that the safety (tissue heating) issue also benefits from my approach, and validate the approach using scanner experiments. Beyond reducing sensitivity by introducing motion-robustness, a second focus of my research contributes towards a real-time pTx pulse design method. This would allow real-time updates to be applied to the scanner, compensating for patient motion as it occurs throughout the scan, and therefore more accurately and effectively negating all motion-induced effects. I am using a machine learning approach to train a neural network with simulated training data for development of the approach, before testing on experiments. Implications of my work are widespread. By removing motion-related concerns over pTx, the benefits of UHF MRI (e.g. the higher image resolution) can be applied clinically. Areas such as epilepsy and MS diagnosis and prognosis have already demonstrated benefits of this in research contexts, but the technical challenges mentioned here currently prevent its widespread or clinical use