Topology-driven Analysis of Brain Anatomy

描述（由申请人提供）：本提案旨在通过约翰霍普金斯大学神经科学教学培训和医学图像处理研究，培养候选人在神经影像学方面进行多学科研究的能力。该研究项目包括创建一个自动化的，定量处理框架的结构脑MRI基于使用拓扑作为一种手段来引入全局，定性解剖信息。它建议开发新的脑图像分割技术，包括多模态研究、病理学和DTI处理，建立人类大脑的拓扑图谱，开发形状分析和同态图像校准工具，并研究MDMA（摇头丸）使用者和多发性硬化症（MS）患者的结构变化。该分割框架将结合拓扑保持分割、组织分类、统计地图集和张量计算的最新进展，自动分割大脑、病变和白质纤维束的主要结构。大脑拓扑图谱将在神经解剖学训练、医学图谱和注释数据库的基础上对大脑的结构和连接进行建模。该分析将结合形状测量和微分形态登记，为两项互补研究提供整体和局部结构量化：一项是对MDMA药物使用者大脑细微形态变化的研究，另一项是对不同程度病变和萎缩的多发性硬化症患者的纵向研究。一旦完全验证，这些工具将免费提供给神经科学研究人员。

DESCRIPTION (provided by applicant): This proposal aims at developing the candidate's ability to perform multidisciplinary research in Neuroimaging through didactic training in Neuroscience and research in Medical Image Processing at the Johns Hopkins University. The research project consists in creating an automated, quantitative processing framework for structural brain MRI based on the use of topology as a means to introduce global, qualitative anatomical information. It proposes to develop novel brain image segmentation techniques encompassing multi-modality studies, pathologies, and the processing of DTI, to build topological atlases of the human brain, to develop tools for shape analysis and homeomorphic image alignment, and to study structural variations in MDMA (ecstasy) users and Multiple Sclerosis (MS) patients. The segmentation framework will combine recent advances in topology-preserving segmentation, tissue classification, statistical atlases and tensor computing to automatically segment the major structures of the brain, lesions, and white matter fiber bundles. Atlases of brain topology will model structure and connectivity in the brain based on neuroanatomy training, medical atlases and annotated databases. The analysis will combine shape measurements with diffeomorphic registration to provide global and local quantification of structure in two complementary studies: a study of subtle morphometric changes in the brain of users of the drug MDMA, and a longitudinal study of Multiple Sclerosis patients with various levels of lesion and atrophy. The tools will become freely available to neuroscience researchers once fully validated. PUBLIC HEALTH RELEVANCE: The impact of the proposed research in the neuroscience and medical communities will be the release of software to segment and align brain MR images with no manual interaction for a large variety of imaging studies, the modeling of structure and connectivity relationships in brain anatomy from topological properties, and the application to two important questions of neurosciences, the detection and quantification of atrophy and the analysis of lesions in the human brain.