EAGER: Geometric Mapping and Diffusion for 3D Imaging Informatics

在绘制和匹配大量多模态跨主题数据的既定软件方法正在成为整个工作流程的瓶颈，并阻碍了对scientiÞc发现的大规模数据的理解和利用的进展。在许多情况下，嵌入在现实世界物体的三维成像中的固有几何结构在映射单个物体以解释它们的相似性和差异性方面非常有效。但最近的计算技术仍然集中在提取和测量单个数据集中的几何和物理属性。几何数据的全球建模以及大型单个数据集内部和之间相关信息的模式和关系的评估尚未得到充分探索。一个严格的计算框架将几何映射和匹配紧密耦合在一起，对于完成大规模图像数据集中包含的各种特征之间的潜在关系的综合分析，并从本质上推进三维成像信息学具有重要意义。本EAGER提案侧重于在黎曼几何映射和几何扩散中潜在的变革性研究思想和方法，这些思想和方法紧密耦合在一起，以建立跨大量学科的精确映射和匹配

Established software approaches in mapping and matching of a large collection of multimodality cross-subject data is becoming a bottleneck for the overall work stream and hindering progress in understanding and utilization of large-scale data for scientiÞc discovery. In many scenarios, intrinsic geometric structures embedded in 3D imaging of real-world objects are very effective in mapping individual objects for interpretation of their similarity and disparity. But recent computational techniques are still focused on the extraction and measurement of geometric and physical properties in a single dataset. Global modeling of geometric data and assessment of patterns and relationships of related information within and across large individual datasets are under-explored. A rigorous computational framework that tightly couples geometric mapping and matching is of great importance to accomplish integrative analysis of a variety of underlying relationships in features of contained in large-scale image datasets and advance 3D imaging informatics substantially. This EAGER proposal focuses on potentially transformative research ideas and approaches in Riemannian geometry mapping and geometric diffusion, which are tightly coupled together to establish the accurate mapping and matching across a large number of subjects