Computational platform for integrated multimodal analysis of brain networks across resolution scales

所有的行为和想法都是通过大脑中的复杂网络实现的。绘制这些网络的连接图对于了解大脑如何处理信息以及这些网络是如何被破坏并导致疾病至关重要。这一困难的映射需要计算工具来自动从多种成像方式(如显微镜数据和脑扫描)中提取信息，并将微观细胞过程与整个大脑的宏观连接结合起来。然而，目前的分析方法受到几个限制，包括不能产生最佳结果，依赖代价高昂的人工干预，或者计算代价高昂。为了抓住这些机会，我们的总体目标是建立一个研究计划，用于自动绘制大脑连接图并对这些多模式数据进行综合分析。我们将开发新的人工智能(AI)算法，在多个测量水平上对这些大脑连接数据进行精确的分析和建模。使用老鼠模型，我们将可视化、提取和模拟关键大脑区域和整个大脑中的细胞到细胞的投影和连接。建立拟议的基于人工智能的平台将满足该领域对全脑层面自动和准确绘制大脑网络图的迫切需求。这项研究计划将开发和优化稳健的人工智能模型，旨在分析和建模大型3D显微镜和脑成像数据。我们的计算模型有可能在细胞水平上增强我们对大脑连通性的理解。重要的是，这些工具可以支持许多其他神经科学、成像或网络科学应用，从而推动生物信息学研究的广泛领域。

All behaviour and thoughts are made possible by complex networks within the brain. Mapping the connections of these networks is essential for understanding how the brain processes information and how these networks are disrupted and contribute to disease. This difficult mapping requires computational tools to automatically distill information from multiple imaging modalities such as microscopy data and brain scans, and integrate microscopic cellular processes with macroscopic connectivity across the whole brain. Current analysis approaches, however, suffer from several limitations including not producing optimal results, relying on costly human intervention or being computationally expensive. To address these opportunities, our overall goal is to build a research program for automated mapping of brain connectivity and integrated analysis of this multimodal data. We will develop novel artificial intelligence (AI) algorithms to perform precise analysis and modelling of this brain connectivity data at multiple levels of measurement. Using mouse models, we will visualize, extract and model the cell-to-cell projections and connections in key brain regions and across the whole brain. Establishing the proposed AI-based platform will address a pressing need in the field for automated and accurate mapping of brain networks at the whole-brain level. This research program will develop and optimize robust AI models designed for the analysis and modeling of large 3D microscopy and brain imaging data. Our computational models have the potential to enhance our understanding of brain connectivity at the cellular level. Importantly, these tools can enable many other neuroscience, imaging or network science applications, thereby advancing the broad field of bioinformatics research.