Multimodality image-based assessment system for traumatic brain injury

描述(由申请人提供)：每年有近170万美国人遭受创伤性脑损伤(TBI)，这构成了美国一个重要的医疗健康问题。尽管神经影像在病理定位和手术计划中起着重要的作用，但目前脑外伤的临床护理并没有充分利用神经影像计算技术。我们建议开发和验证基于图像分割、配准和分析的计算算法，这些算法产生量化措施来表征损伤、监控病理演变、告知患者预后和优化患者护理工作流程。该项目解决了目前临床对信息丰富的脑损伤指标的需求，以及对能够处理导致严重脑变形的大型、异质病理的易于使用的图像分析工具的技术需求。在目标1中，我们将执行多模式脑图像分割，以评估急性和慢性脑外伤，并测量纵向变化。我们将生成基于从多模式图像数据集中分割病变、出血、脑室、灰质(GM)、白质(WM)和大脑中线的脑损伤病理的定量测量。临床上，这些指标将用于定量描述和评估任何时间点(急性、慢性)的损伤，并根据病理类型、位置和程度进行纵向跟踪。该项目的第二个目标是推进图像配准的最新技术，用于急性和慢性脑损伤的评估和纵向变化的测量。可变形图像配准在图像中对齐相应的解剖结构，并返回封装它们之间的变形的位移或流场。我们将继续开发“几何变形”，可以配准由结构生长或收缩引起的外观显著变化的图像，如脑外伤病理。我们将获得新的体素量化和可视化的病理渗透和由损伤或纵向脑变化引起的脑变形，这两者都在 病变以及GM和WM内。第三个目标是研究我们从图像分割和配准中衍生出来的新的TBI指标预测结果和指导临床决策的能力。重点放在最终的临床影响上，以及评估大脑重塑(例如结构变化)与功能恢复或衰退之间的关系。我们将使用多元统计方法来评估AIMS 1-2中基于图像的新的多模式脑损伤测量(体积和基于变形的测量)相对于每个脑外伤患者可用的神经心理运动、认知和行为结果测量的预后能力。多变量技术还将允许对患者分组进行调查 基于描述其共性或以最佳方式区分它们的统计特征的群体。这将有助于针对每个患者分组定制特定的临床工作流程。最终，这里提出的技术进步将产生使用成像以综合的、纵向的方式监测大脑对创伤的反应的能力，具有最大的临床实用性和特异性。

DESCRIPTION (provided by applicant): Nearly 1.7 million Americans suffer traumatic brain injury (TBI) annually, which constitutes an important and significant US medical health concern. Although neuroimaging plays an important role in pathology localization and surgical planning, TBI clinical care does not currently take full advantage of neuroimaging computational technology. We propose to develop and validate computational algorithms, based on image segmentation, registration and analysis, which yield quantitative measures to characterize injury, monitor pathology evolution, inform patient prognosis and optimize patient care workflows. This project addresses the current clinical need for informative TBI metrics and the technical need for easy-to-use image analysis tools capable of handling large, heterogenous pathologies that cause severe brain deformations. In Aim 1, we will perform multimodal brain image segmentation for the assessment of acute and chronic TBI, and for measuring longitudinal changes. We will generate quantitative measures of TBI pathology that are based on segmenting lesions, hemorrhages, ventricles, gray matter (GM), white matter (WM) and the brain midline from multimodal image datasets. Clinically, these metrics will be used to quantitatively describe and assess injury at any time point (acute, chronic) and for longitudinal tracking based on pathology type, location and extent. The second aim of this project is to advance the state-of-the-art in image registration for acute and chronic assessment of TBI and for longitudinal change measurement. Deformable image registration aligns corresponding anatomy in images and returns a displacement or flow field encapsulating the deformations between them. We will continue development of "geometric metamorphosis", can register images with significant appearance changes caused by structures that grow or contract, such as TBI pathologies. We will derive novel voxel-wise quantifications and visualizations of pathology infiltration and of brain deformations induced by injury or longitudinal brain changes, both within lesions and within GM and WM. The third aim is to investigate the ability of our novel TBI metrics, derived from image segmentation and registration, to predict outcome and guide clinical decision making. The focus is on final clinical impact and on evaluating the relationship between brain remodeling (e.g. structural changes) with functional recovery or decline. We will use multivariate statistical methods to evaluate the prognostic abilities of the novel multimodal image-based measures of TBI (volumetric and deformation-based) from Aims 1-2 with respect to the neuropsychological motor, cognitive and behavioral outcome measures available for each TBI patient. Multivariate techniques will also allow investigation into the grouping of patient sub populations based on statistical features that describe their commonalities or optimally differentiate between them. This will aid in the customization of clinical workflows specific to each patient sub- group. Ultimately, the technical advances being proposed here will yield the ability to use imaging to monitor brain responses to trauma in an integrative, longitudinal fashion, with maximal clinical utility and specificity.