An Alignment Framework For Mapping Brain Dynamics and Substrates of Human Cognition Across Species

摘要 非人灵长类动物（NHP）模型对于转化神经科学的发展至关重要，因为它允许 研究人员将有关人类宏观大脑动力学和认知的观察与潜在的 在人类中无法完全研究的中尺度和微观尺度现象。重要的是，最终价值 NHP的研究结果为人类模型提供信息依赖于跨物种方法的充分性 解剖和功能对齐。在这方面，基于解剖学标志的跨物种大脑方法 排列在低级感觉和运动区表现出色，但在记录异模态方面面临限制 关联区域，其具有缺乏地标。作为回应，我们开发了一种基于fMRI的交叉- 物种对齐框架，利用网络组织表示的最新进展， 生成公共坐标空间（称为“联合嵌入”）。我们对这个函数的最初应用- 基于方法的皮质对齐使我们能够量化人类和猕猴之间的高同源性， 所有皮层层次的细节。在这个建议中，我们将扩展联合嵌入对齐 使用多模态MRI数据在人类和非人类灵长类动物之间对齐大脑的方法（目标1）。我们将 利用大样本多模态MRI数据集的公开可用性（人类连接组项目 [HCP]，可靠性和生殖联盟[CoRR]），以及最近公开分享的非人类 灵长类动物数据（PRIMatE数据交换），以纳入物种内和物种之间的变化。我们将评估 通过与传统的基于地标的配准和单峰联合配准的比较， 嵌入对齐。使用最高性能的对齐来转换物种之间的大脑地图，我们 将量化人类和猕猴之间大脑网络的时空相似性和差异 基于自回归、准周期模式和共激活模式分析（目标2）。此外，我们将 将人类认知本体映射转换到猕猴空间，并评估对应的 人类和猕猴之间每个认知成分的大脑网络。我们还将建立一个互动的 网页浏览器分享翻译人类认知本体在猕猴（目的3）。所有数据、数据产品 (e.g.跨物种翻译）和生成的代码将通过开放科学资源公开共享- PRIME-DE。

ABSTRACT The non-human primate (NHP) model is critical to the advancement of translational neuroscience, as it allows researchers to link observations regarding macroscale brain dynamics and cognition in the human to underlying meso- and microscale phenomena that cannot be fully investigated in humans. Importantly, the ultimate value of findings from the NHP for informing human models relies on the adequacy of methods for cross-species anatomical and functional alignment. In this regard, anatomical landmark-based methods for interspecies brain alignment have excelled in lower order sensory and motor areas, but faced limitation in registering heteromodal association areas, which have a paucity of landmarks. In response, we have developed an fMRI-based cross- species alignment framework that leverages recent advances in representation of network organization to generate a common coordinate space (referred to as “joint-embedding”). Our initial application of this function- based method for cortical alignment allowed us to quantify homologies between human and macaque in high detail at all levels of the cortical hierarchy. In this proposal, we will extend the joint-embedding alignment approach to align brain between human and non-human primates using multimodal MRI data (Aim 1). We will make use of the publicly availability of large sample multimodal MRI datasets (Human Connectome Project [HCP], Consortium for Reliability and Reproducibility [CoRR]), as well as recent openly shared non-human primate data (PRIMatE Data Exchange) to incorporate of within- and between species variations. We will assess the alignment performance by comparing to traditional landmark-based registration and unimodal joint- embedding alignments. Using the highest performing alignment to transform brain maps between species, we will quantify spatiotemporal similarities and divergence of brain network between human and macaque monkey based on autoregressive, quasi-periodic pattern, and coactivation pattern analysis (Aim 2). Additionally, we will transform the human cognitive ontology maps to macaque space and assess the similarities of corresponding brain networks for each cognitive component between human and macaque. We will also build an interactive webpage viewer to share the translation human cognition ontology in macaque (Aim 3). All data, data products (e.g. cross-species translation) and code generated will be openly shared through the open science resource - PRIME-DE.