A novel multi-modal, multi-scale imaging pipeline for the validation of diffusion MRI of the brain

项目摘要 在这个项目中，我们建议从扩散张量图像中验证和表征纤维方向估计。 通过优化全小鼠大脑的多模态、多尺度成像管道，使用DTI技术。DTI 是一个强大的工具，用于非侵入性报告3D微观结构的神经组织的宏观特性， 量表，并在许多神经系统疾病的理解和诊断中发挥了重要作用 流程.可以处理DTI数据的现代采集，以生成称为 取向扩散函数（ODF）。ODF用于推断局部轴突纤维的方向， ulations。以前验证这些方向估计的努力主要依赖于连续光学组织学 作为地面实况数据集。基于组织学的管道涉及物理切片的劳动密集型任务， 将组织切割成薄片，导致样品的物理破坏和各向异性分辨率。这些限制 潜在地混淆3D取向估计的准确性，使空间配准的过程复杂化， 地面实况和DTI数据集，并限制定量比较，以选择整个感兴趣区域（ROI） 大脑样本 近年来，同步加速器X射线显微计算机断层扫描（microCT）已成为一种强大的工具， 高分辨率组织成像利用镶嵌投影拼接方法，可以对整个小鼠大脑进行成像 在先前用DTI成像之后，以1.2微米的各向同性3D分辨率。为了增强microCT对比度，组织- 将样本固定并用电子显微镜（EM）中使用的相同类型的金属基染色剂染色 在嵌入树脂之前。我们将优化该microCT-EM验证管道，以解决预 基于组织学的研究，并使用 微米到纳米级的神经学信息。 该提案的具体目标是：（1）模型相位对比，以优化microCT数据采集，（2）vali， 最新的DTI ODF重建方法，使用地面实况microCT（3）表征DTI性能， 从EM中提取组织微结构信息。完成后，aim 1将产生一个新的理论模型 和采集策略，以利用强吸收生物样品中的microCT相位对比。目标2将 在整个小鼠大脑中生成ODF的地面实况数据集，该数据集将用于计算算法- DTI性能的特定空间图。在目标3中，将选择大约20个ROI进行纳米级成像， EM和DTI性能将通过潜在神经结构的定量特征来表征。 这些结果将提供一个前所未有的微结构驱动的理解的DTI信号，使未来， 真实研究，开发更先进的DTI模型和采集策略，以更好地利用纤维定向 和连接信息在神经疾病治疗中的应用。

Project Summary In this project, we propose to validate and characterize fiber orientation estimation from diffusion tensor imag- ing (DTI) through the optimization of a multi-modality, multi-scale imaging pipeline for whole mouse brains. DTI is a powerful tool used to noninvasively report 3D microstructural properties of nervous tissue on a macroscopic scale, and has played an important role in the understanding and diagnosis of a number of neurological disease processes. Modern acquisitions of DTI data can be processed to generate a 3D diffusion profile known as an orientation diffusion function (ODF) at each voxel. The ODF is used to infer the orientation of local axon fiber pop- ulations. Previous efforts to validate these orientation estimates have primarily relied on serial optical histology as a ground truth dataset. Histology-based pipelines involve the labor intensive task of physically sectioning the tissue into thin slices, leading to physical destruction of the sample and anisotropic resolution. These limitations potentially confound the accuracy of 3D orientation estimation, complicate the process of spatially registering the ground-truth and DTI datasets, and limit quantitative comparisons to select regions of interest (ROI) across the brain sample. In recent years, synchrotron x-ray microcomputed tomography (microCT) has emerged as a powerful tool for high-resolution tissue imaging. With a mosaic projection-stitching method, a whole mouse brain can be imaged at an isotropic, 3D resolution of 1.2 microns after prior imaging with DTI. To enhance microCT contrast, the tis- sue specimens are fixed and stained with the same kind of metal-based stains used in electron microscopy (EM) prior to embedding in resin. We will optimize this microCT-EM validation pipeline to address the limitations of pre- vious histology-based studies, and characterize DTI algorithm performance across a whole mouse brain using micron- to nano-scale neurological information. The specific aims of the proposal are: (1) model phase contrast to optimize microCT data acquisition, (2) vali- date DTI ODF reconstruction methods using ground-truth microCT (3) characterize DTI performance using under- lying tissue microstructure information from EM. Upon completion, aim 1 will generate a novel theoretical model and acquisition strategy to exploit microCT phase contrast in strongly absorbing biological samples. Aim 2 will generate a ground-truth dataset of ODFs across a whole mouse brain, which will be used to calculate algorithm- specific spatial maps of DTI performance. In Aim 3, around 20 ROI will be selected for nano-scale imaging with EM, and DTI performance will be characterized by quantitative features of the underlying neural architecture. These results will provide an unprecedented microstructure-driven understanding of the DTI signal, allowing fu- ture studies to develop more advanced DTI models and acquisition strategies to better leverage fiber orientation and connectivity information in the treatment of neurological disease.