Multicomponent Modeling of High-Dimensional Multiparametric MRI Data

项目总结 高维多参数磁共振数据的多分量建模 核磁共振成像产生毫米级空间分辨率的图像，同时出现许多重要的生物特征 在更小的(微观)尺度上。在过去的几十年里，MRI从业者使用了 从生物物理参数(如松弛和扩散)获得的信息间接探测微观 使用毫米级数据的组织隔间。虽然这些方法在一定程度上取得了成功，但 近年来出现了利用多参数MRI数据的创新新范例(例如，使用 在高维实验中联合松弛和扩散)来探测组织微结构 前所未有的详细程度。尽管这种多分量多参数方法可能相当强大， 图像采集和分析方法的实质性改进以及可访问性的改进是 需要这些方法才能被更广泛的社区常规用于实际应用。这个 拟议的项目涉及开发新的分析方法来识别和分离多个 使用高级约束估计技术从MRI数据中提取微结构组织分区， 开发了一种新型的端到端图像预处理流水线(包括配准、失真等步骤 校正、去噪等)这是专门为多参数采集而设计的，开发 评估估计质量和优化多参数采集协议的新工具，以及 将这一方法应用于体外培养的小鼠大脑和脊髓，为性行为提供新的见解 氧化应激在多发性硬化症小鼠模型中的差异和作用。此外，新的方法 我们开发的将集成到开源BrainSuite扩散管道软件包中，该软件包将， 第一次为更广泛的成像社区提供对这些强大方法的轻松访问。

PROJECT SUMMARY Multicomponent Modeling of High-Dimensional Multiparametric MRI Data MRI generates images with millimeter-scale spatial resolution, while many important biological features occur at much smaller (microscopic) scales. Over the past several decades, MRI practitioners have used the information derived from biophysical parameters (like relaxation and diffusion) to indirectly probe microscopic tissue compartments using millimeter-scale data. While these approaches have been somewhat successful, an innovative new paradigm has emerged in recent years that leverages multiparametric MRI data (e.g., using relaxation and diffusion jointly in a higher-dimensional experiment) to probe tissue microstructure with an unprecedented level of detail. Although such multicomponent multiparametric methods can be quite powerful, substantial improvements in image acquisition and analysis methods and improvements in accessibility are needed for these approaches to be used routinely for practical applications by the broader community. The proposed project involves the development of novel analysis methods to identify and separate multiple microstructural tissue compartments from MRI data using advanced constrained estimation techniques, the development of a novel end-to-end image preprocessing pipeline (including steps like registration, distortion correction, denoising, etc.) that is especially designed for multiparametric acquisitions, the development of novel tools to evaluate estimation quality and optimize multiparametric acquisition protocols, and the application of this approach to ex vivo mouse brains and spinal cords to provide new insights into sex differences and the role of oxidative stress in a mouse model of multiple sclerosis. Further, the new methods we develop will be integrated into the open-source BrainSuite Diffusion Pipeline software package, which will, for the first time, provide the broader imaging community with easy access to these powerful approaches.