Design and optimization of medical devices for MRI-guided radiotherapy applications

预计到2020年，加拿大将有超过20万例新发癌症病例，其中近一半的病例中，放射治疗（RT）在疾病管理中发挥着关键作用。放射治疗是在用足够致死剂量的辐射准确治疗目标肿瘤与尽量减少对周围健康组织的剂量之间取得平衡。图像引导RT极大地提高了我们使治疗日益适形的能力，提高了治疗比例，并使治疗适应治疗过程中肿瘤形状和大小的变化。医学线性加速器的进步已经提供了毫米级的剂量传递精度：成像系统的性能现在是适形剂量传递和提供改进的护理的限制因素。磁共振成像提供了优越的软组织对比，更好地区分健康组织和癌性肿瘤，并越来越多地用于诊断，治疗计划和最近的室内指导RT治疗交付。最近，新颖的MR-Linac技术已经建立在这个概念上。然而，核磁共振是一种复杂的成像方式，需要先进的技术和临床专业知识来确保其有效性和安全性。人们认识到，由于扫描仪规格和性能以及患者诱导的磁场效应，MR图像本质上遭受几何扭曲。我们的计划旨在通过定义新的理论和实验方法，通过开发新一代MR图像质量（MR- iq）量化设备，减少RT规划和指导中的这些不确定性。该研究计划的目标是：(1)开发基本方法，设计和测试MRID3D扩展视场，使用谐波分析和磁化率模拟；(2)利用IQ插件对MRID3D进行建模，设计出一套综合的MR-IQ管理装置；(3)开发基于机器学习方法的自动化MR-IQ图像处理流水线。该计划的影响是提供总体上更好的一致性，概念清晰度，并提高了MR-IQ与放射治疗应用相关的准确性。从长远来看，全面MR- iq的效率和准确性的提高将导致MR技术在RT中的更广泛接受和更深层次的集成，并将隐含地促进更好的结果。我们的项目建立在改变MR-IQ实践的记录之上，我们的一些技术可用于全球最终用户的测试和临床实施。

Over 200,000 new cases of cancer are expected in Canada in 2020 with radiation therapy (RT) playing a pivotal role in the management of the disease in nearly half of the cases. RT is a balance between accurately treating the target tumour with sufficiently lethal dose of radiation while minimizing the dose to surrounding healthy tissue. Image-guided RT greatly improves our ability to make treatments increasingly conformal, improving this therapeutic ratio, and adapting treatment to changes in tumour shape and size over the course of treatment. Advances in medical linear accelerators have provided mm-level accuracies in delivering dose: imaging system performance is now the limiting factor in conformal dose delivery and providing improved care. Magnetic resonance (MR) imaging provides superior soft-tissue contrast to better distinguish healthy tissue from cancerous tumours and has been increasingly used for diagnosis, treatment planning and more recently in-room guidance for RT treatment delivery. Recently, novel MR-Linac technologies have been built on this concept. However, MR is a complex imaging modality that requires both advanced technical and clinical expertise to ensure its effectiveness and safety. It is recognized that MR images intrinsically suffer of geometric distortions due to both scanner specifications and performance as well as patient-induced magnetic field effects. Our program aims to reduce these uncertainties in RT planning and guidance by developing new-generation devices for the quantification of MR image quality (MR-IQ), by defining new theoretical and experimental methodologies. The aims of the research program are (1) to develop fundamental methods, design, and test MRID3D with extended field of view using harmonic analysis and magnetic susceptibility simulations; (2) to model MRID3D with IQ inserts to design a comprehensive device for the management of MR-IQ; (3) to develop an automated MR-IQ image processing pipeline powered by machine learning methodology. The impact of this program is to provide overall better consistency, conceptual clarity, and enhanced accuracy to the MR-IQ relevant to radiation therapy applications. Improved efficiency and accuracy for comprehensive MR-IQ in the long term will lead to a wider acceptance and deeper integration of MR technologies in RT, and implicitly will contribute to better outcomes. Our program builds on a track record of practice-changing MR-IQ with some of our technologies available to end-users worldwide for testing and clinical implementation.