Heterogeneity of Multi-modal Imaging Signatures of Aging, MCI, Alzheimer's disease via Pattern Analysis

摘要 阿尔茨海默病(AD)及其前驱阶段对社区和 医疗保健系统。其他几个病理过程常常与AD病理共存。描述了 衰老和(前驱)AD中存在的大脑变化的多方面方面不仅提供了对 潜在的病理生理过程，如淀粉样沉积、小血管缺血性疾病(SVID)、 和神经退行性变或功能改变，但也导致这些病理过程的生物标志物。在……里面 这一建议，我们的目标是捕捉反映病理过程的各种成像模式的异质性 通过应用最先进的异质性分析机器学习，发生在衰老和前驱AD 方法本实验室新近发展起来的多角度成像数据(结构磁共振、静息状态 功能磁共振成像、淀粉样蛋白成像)。这些方法使我们能够量化不同的方式/模式 个体可能会脱离标准的大脑老化轨迹。我们的目标是实现一种新的基于图像的 老龄化和神经退行性疾病协调系统“(ISTAGING)，其中每个方面都将 反映大脑变化的不同模式，从而捕捉潜在的神经解剖学、神经功能 以及可量化和可复制指标中的神经病理异质性。此外，我们的目标是将这些 神经成像表型与神经认知表型以及认知正常进展的关系 衰老到MCI和痴呆症。这将允许我们将每个个体放入一个新的维度大脑中 老化的坐标系和他/她的轨迹图，以及确定从 多参数成像数据。在实现这些目标之前，我们将加强和应用各种成像 协调方法，这是确保跨研究和扫描仪的数据可以整合的关键 以建设性的方式。特别是，Aim1将应用站点间成像协调方法，因此 开发一个平台，使我们能够集成10,000多个多参数成像数据集 几项研究的参与者，年龄跨度从45岁到89岁；因此，我们将形成一个大型多模式成像 数据库捕捉正常脑老化和前驱AD的异质性。AIM 2将使用机器 整合脑萎缩、SVID、Aβ沉积和功能连接的空间模式的学习方法 在这个人群中，并将得出标准的脑老化曲线。目标3将使用最先进的半监督 区分恢复期和进展期成像模式异质性的学习方法 大脑老化：这将导致维度系统(ISTAGING)捕获不同的维度/模式 与恢复力大脑老化相关的高级(非恢复力)大脑变化，反映了潜在的 神经病理过程。目标4将使iSTAGING坐标和纵向轨迹与 认知、危险因素和从正常认知到MCI再到痴呆的临床进展，以便进一步 阐明这些神经影像维度的临床相关性。

Abstract Alzheimer's Disease (AD), as well as its prodromal stages, poses a significant burden to the community and health care system. Several other pathologic processes often co-exist with AD pathology. Characterizing the multi-faceted aspects of brain changes present in aging and (prodromal) AD not only provides insights into the underlying pathophysiological processes, such as amyloid deposition, small vessel ischemic disease(SVID), and neurodegeneration or functional change, but also leads to biomarkers of these pathologic processes. In this proposal, we aim to capture the heterogeneity of various imaging patterns that reflect pathologic processes occurring with aging and prodromal AD, by applying state of the art heterogeneity analysis machine learning methods recently developed in our laboratory to multi-faceted imaging data (structural MRI, resting state functional MRI, amyloid imaging). These methods allow us to quantify the different ways/patterns in which an individual can fall off the normative brain aging trajectories. Our goal is to arrive at a new “Imaging-based coordinate SysTem for AGing and NeurodeGenerative diseases”(iSTAGING), each dimension of which will reflect a different pattern of brain alterations, hence capturing the underlying neuroanatomical, neurofunctional and neuropathological heterogeneity in quantifiable and replicable metrics. Moreover, we aim to link these neuroimaging phenotypes with neurocognitive phenotypes, as well as with progression from cognitively normal aging to MCI and to dementia. This will allow us to place each individual into a new dimensional brain coordinate system of aging and map his/her trajectory, as well to determine predictive indices emanating from multi-parametric imaging data. Prior to achieving these aims, we will strengthen and apply various imaging harmonization methods, which are critical in ensuring that data across studies and scanners can be integrated in a constructive way. In particular, Aim1 will apply inter-site imaging harmonization methods, and hence develop a platform that will allow us to integrate multi-parametric imaging datasets from more than 10,000 participants to several studies spanning ages 45 to 89; we will therefore form a large multi-modal imaging database capturing the heterogeneity of normal brain aging and prodromal AD. Aim 2 will use machine learning methods to integrate spatial patterns of brain atrophy, SVID, Aβ deposition and functional connectivity in this population, and will derive normative brain aging curves. Aim 3 will use state of the art semi-supervised learning methods to disentangle the heterogeneity of imaging patterns distinguishing resilient from advanced brain aging: this will lead to the dimensional system (iSTAGING) capturing the diverse dimensions/patterns of advanced (non-resilient) age-related brain change relative to resilient brain aging, which reflect underlying neuropathological processes. Aim 4 will relate iSTAGING coordinates and longitudinal trajectories to cognition, risk factors, and clinical progression from normal cognition to MCI to dementia, in order to further elucidate the clinical correlates of these neuroimaging dimensions.