Quantifying image distortion in MRI for Radiotherapy treatment planning

核磁共振成像是一种重要的医学成像方式。它提供具有独特软组织对比度的高质量3D图像，在包括肿瘤成像和手术计划在内的广泛临床应用中非常有用。一个关键的应用是放射治疗，MRI越来越多地用于计划向肿瘤部位放射，同时尽量减少对邻近健康组织的剂量。与使用CT的传统方法相比，MRI为靶向和治疗监测提供了更好的组织对比和定量成像生物标志物定义（使肿瘤更容易描绘），并且还消除了在成像期间将患者暴露于电离辐射的需要。然而，一个非常重要的考虑因素是MR图像中的空间失真。患者在扫描仪中的存在会扭曲外加磁场，这意味着图像在组织特征的大小、形状和位置上包含小误差。这意味着辐射传递中的误差和不确定性不能完全量化，这意味着必须使用非常保守的估计，而不是在治疗计划管道的其他地方应用的谨慎方法。本博士旨在详细了解MR图像失真，以充分量化其影响并将其纳入治疗计划管道。该项目部分是基于精心构建和成像测试对象（也称为幻象），这些对象以受控的方式扭曲了场，从简单，易于理解的对象构建到更复杂的对象，这些对象更能代表器官（如大脑）的几何和生物标志物特征，部分是基于对图像失真背后的物理建模。目的是定量表征畸变场，并将其纳入放射治疗审计。它的范围包括基于物理的建模、统计和其他先进方法，如人工智能，以尽可能完整和准确地描述图像失真，为治疗计划和监测提供信息。监督小组由核磁共振、建模和测量科学方面的专家组成。它与一家领先的MRI幻影制造商合作，后者带来了先进材料和制造能力的知识。我们还与在英国全国范围内提供放射治疗审计的团队密切合作。成功的申请者将获得实际成像、高级建模、图像分析和测量科学方面的经验，并有机会在完成工作时强调自己的优势和兴趣。该项目将包括：1；开发敏感性匹配材料，包括可制造性和制造过程中相关不确定性的评估。一系列复杂的3D打印变形模型的设计和表征，包括关键物理特性的模拟。测试对象和志愿者的核磁共振扫描幻影与应用场与脉冲序列的模拟评估图像失真校正算法的性能，量化重建中的误差6。由场畸变效应和处理引起的基于图像的不确定性的量化

MRI is a key medical imaging modality. It provides high quality 3D images with unique soft tissue contrast which are useful across a huge range of clinical applications including imaging tumours and for surgical planning. One key application is in radiotherapy, where MRI is increasingly used to plan the delivery of radiation to the site of a tumour while minimising the dose to neighbouring, healthy tissue. Compared to more traditional approaches using CT, MRI provides improved tissue contrast and quantitative imaging biomarker definition (making the tumour easier to delineate) for targeting and treatment monitoring, and also removes the need to expose the patient to Ionising radiation during imaging.One very important consideration, however, is spatial distortion in the MR images. The presence of a patient in the scanner distorts the applied magnetic field and means that images contain small errors in the size, shape, and position of tissue features. This means that the errors and uncertainties in radiation delivery cannot be fully quantified, meaning that very conservative estimates must be used instead of the careful approaches applied elsewhere in the treatment planning pipeline.This PhD aims to develop a detailed understanding of MR image distortion to fully quantify its effects and incorporate them into the treatment planning pipeline. The project is partly based on carefully building and imaging test objects (also known as phantoms) which distort the field in a controlled way, building from simple, easily understood objects to more complex objects which are more representative of the geometry and biomarker characteristics of organs such as the brain, and partly on modelling the physics behind image distortion. The aim is to quantitatively characterise the distortion field and to incorporate this into Radiotherapy audits. There is scope to include physics-based modelling, statistics, and other advanced approaches such as AI to inform treatment planning and monitoring with the most complete and accurate description of image distortion possible.The supervision team comprises experts in MRI, modelling, and measurement science. It is in partnership with a leading MRI phantom manufacturer who bring knowledge of advanced materials and manufacturing capabilities. We also work closely with teams who provide radiotherapy audits nationally across the UK. The successful applicant would gain experience across practical imaging, advanced modelling, image analysis, and measurement science and would have the opportunity to emphasize their own strengths and interests in completing the work.The project will cover:1. Development of susceptibility-matched materials, including evaluation of manufacturability and associated uncertainties in manufacturing2. Design and characterization of 3D printed distortion phantoms of a range of complexities Including mimics of key physical properties3. MRI scanning of test objects and volunteers4. Simulation of phantoms and applied field and pulse sequences5. Assessing the performance of image distortion correction algorithms, quantifying errors in reconstruction6. Quantification of image-based uncertainties due to field distortion effects and processing