Quasi-rigid image registration for DCE-MRI

描述（由申请人提供）：动态对比增强磁共振成像（DCE-MRI）生物标志物显示出作为现代临床治疗反应早期指标的重要前景。因此，他们拥有一个不断发展的作用，在当前的个性化医疗领域。然而，关于它们在横截面同期或纵向研究中的应用仍然存在挑战-在这些研究中，将由于扫描内患者运动或在扫描仪上重新定位而引起的误差与由于这些生物标志物反映的治疗而发生的解剖或功能变化解耦是至关重要的。因此，迫切需要能够快速但可靠地导出DCE-MRI生物标志物的工具。这项拟议工作的目标是开发、实施和验证三维图像配准技术，以便从临床DCE-MRI中快速推导出疾病相关生物标志物。重点是涉及刚性（骨）和非刚性（软组织）组件的身体区域的受试者内解剖图像的对齐。在我们的创新配准方案中，预先确定的组件（如骨骼）以刚性方式变形，而软组织则以非刚性方式变形。因此，我们假设，与目前临床使用的技术以及最先进的配准方法相比，我们的方法将提供更好的患者重新定位建模。在我们的特定目标中，我们将：（1）开发用于MR图像对齐的准刚性受试者内配准方法，以及（2）在回顾性环境中使用软组织肉瘤成人患者的MR图像验证配准方法。基于计算机和人类观察员将被用来比较所提出的方法对传统的刚性和非刚性配准方法的性能。一个新的系统，放射学评分的优点，图像配准的设计和测试。登记方法将作为一种开放源码技术开发，便于分发。DCE-MRI生物标志物目前正在几种新型药物治疗的临床试验中作为替代终点进行评估。拟定的配准方法将能够在治疗期间对DCE-MRI生物标志物进行快速定量变化分析。医生将能够使用这些信息来根据患者的个人需求定制治疗策略。在这种情况下，本提案中开发的技术适用于广泛的人类疾病，例如，肿瘤学的（例如，肉瘤，头颈癌等）或肌肉骨骼疾病（例如，关节炎等）。这项工作是与Robert Canter密切合作的，Robert Canter是一名外科肿瘤学家，专门从事软组织肉瘤的临床管理，Michael Buonocore是一名MRI物理学家，其专业知识在于临床MRI方案设计和图像分析，Robert Boutin是一名临床放射科医生，擅长软组织肉瘤和顾问韦恩蒙斯基，癌症成像生物标志物开发专家。

DESCRIPTION (provided by applicant): Dynamic-contrast enhanced magnetic resonance imaging (DCE-MRI) biomarkers are displaying significant promise as early indicators of therapeutic response in the modern clinic. As a consequence, they possess an evolving role in the current realm of individualized medicine. Challenges remain however regarding their application in cross-sectional contemporaneous or longitudinal studies - it is critical in these studies to decouple errors due to intra-scan patient motion or repositioning on the scanner from anatomical or functional changes occurring due to treatment that these biomarkers reflect. Tools that are capable of rapid, but reliable derivation of DCE-MRI biomarkers are therefore urgently needed. The GOAL of this proposed work is to develop, implement and validate three-dimensional image registration technology for enabling the rapid derivation of disease-associated biomarkers from clinical DCE-MRI. The focus is on the alignment of intra-subject anatomical images of body regions that involve both rigid (bone) and non-rigid (soft tissue) components. In our innovative registration scheme, pre-determined components such as bones deform in a rigid manner while soft tissue warps non-rigidly. We HYPOTHESIZE therefore that our method will provide better modeling of patient repositioning compared to currently clinically used technologies as well as state-of-the-art registration methods. In our SPECIFIC AIMS, we will: (1) develop the quasi-rigid, intra-subject registration method for MR image alignment, and (2) validate the registration method using MR images of adult patients with soft tissue sarcoma in a retrospective setting. Both computer-based and human observers will be used to compare the performance of the proposed method against conventional rigid and non-rigid registration methods. A novel system for radiologic scoring of the merits of image registration is devised and will be tested. The registration method will be developed as an open-source technique allowing easy distribution. DCE-MRI biomarkers are currently being evaluated as surrogate end-points in clinical trials of several novel drug treatments. The proposed registration method will enable rapid quantitative change analyses of DCE-MRI biomarkers over the treatment period. Physicians will be able to use this information to tailor therapeutic strategies according to the individual requirements of their patients. In this context, techniques developed in this proposal are applicable to a broad class of human diseases, e.g., oncologic (e.g., sarcomas, head and neck cancer, etc) or musculoskeletal disease (e.g., arthritis, etc). This work is a close collaboration with Robert Canter, a surgical oncologist who specializes in the clinical management of soft tissue sarcoma, Michael Buonocore, an MRI physicist whose expertise lies in clinical MRI protocol design and image analysis, Robert Boutin, a clinical radiologist with expertise in soft tissue sarcoma and consultant Wayne Monsky, an expert in cancer imaging biomarker development.