Object-Centric Computational Model for Imaging Analysis

描述（由申请人提供）：本申请提出实施并广泛验证一种新的计算方法，用于使用磁共振成像分析大脑中的解剖结构。 这种称为连续内侧表示的方法是一种创新方法，它使用解剖结构的形状来表征萎缩，并提供一个自然的坐标系，在该坐标系中，来自人群研究中多个受试者的图像信息可以以比以前更高的准确性和更高的统计功效进行组合。 该方法与将大脑视为整体的现有技术有很大不同，这必然限制了对特定感兴趣结构进行分析的灵敏度和特异性。 初步结果提供了令人信服的支持优雅的方法，并为它的潜力进行详细的研究的解剖和功能的影响，疾病和损伤。 该提案的具体目标是：（1）验证cm-rep模型可以准确地表示海马，特别是在存在病理的情况下;（2）通过检查cm-rep方法对齐海马子字段的程度来验证跨学科标准化;（3）确认基于形状的标准化可以提高组fMRI分析的统计功效的初步发现;（4）证明对于海马形状和多变量/多模态成像数据的组合统计分析，CM-REP方法呈现了对基于体积和基于边界的方法的改进。 这些目标将通过将新技术应用于涉及颞叶癫痫、阿尔茨海默病和精神分裂症的功能和结构研究的模拟和真实的成像数据来实现。 这项探索性研究，如果成功的话，将为一些后续的假设驱动的研究，解决结构海马异常和功能招聘的海马在不同的临床人群的记忆任务的研究奠定基础。 这项提案的具体目标可以对健康和疾病中海马体的研究产生直接影响。 海马体的功能涉及编码和巩固新的记忆，在一些最常见的神经系统疾病中具有核心重要性，包括癫痫和痴呆症。 本申请中提出的新方法的精细分析有望通过检测和精确定位海马解剖结构和功能中比当前方法允许的更小，更集中的变化来影响我们对这些破坏性疾病的理解。

DESCRIPTION (provided by applicant): This application proposes to implement and extensively validate a new computational approach for analyzing anatomical structures in the brain using magnetic resonance imaging. This approach, called the continuous medial representation, is an innovative method that uses the shape of anatomical structures to characterize atrophy and to provide a natural coordinate system in which image information from multiple subjects in a population study can be combined with greater accuracy and higher statistical power than previously possible. The approach differs significantly from existing techniques that consider the brain as a whole, which necessarily limits the sensitivity and specificity of analyses on particular structures of interest. Preliminary results provide compelling support for the elegance of the approach and for its potential to conduct detailed studies of both anatomical and functional effects of disease and injury. The specific aims of the proposal are (1) to verify that cm-rep models can represent the hippocampus accurately, particularly in the presence of pathology; (2) to validate cross-subject normalization via the cm-rep method by examining how well it aligns hippocampal sub-fields; (3) to confirm preliminary findings indicating that shape-based normalization can improve the statistical power of group fMRI analysis; (4) to demonstrate that the cm-rep approach presents an improvement over volume-based and boundary-based methods for combined statistical analysis of hippocampal shape and multivariate/multimodal imaging data. These aims will be achieved by applying the new technique to simulated and real imaging data from functional and structural studies involving temporal lobe epilepsy, Alzheimer's disease, and schizophrenia. This exploratory research, if successful, will clear the ground for a number of follow-up hypothesis-driven research studies addressing structural hippocampal abnormalities and functional recruitment of the hippocampus in memory tasks in various clinical populations. The specific aims of this proposal can have an immediate impact on studies of the hippocampus in health and disease. The hippocampus, whose function involves encoding and consolidating new memories, is of central importance in some of the most common neurological disorders, including epilepsy and dementia. Refined analysis with the new approach proposed in this application promises to impact our understanding of these devastating disorders by detecting and pinpointing smaller, more focal changes in hippocampal anatomy and function than the current methods allow.