Computational Architectures and Visualization: Real-time 3D Data Analysis and Tomographic Reconstruction in a Reproducible Visualization Platform
Abstract
This project will define capabilities not seen before in 3D metrology---real-time tomographic reconstructions as data is acquired, 3D chemical visualization (e.g. spectroscopic / multimodal), and a state-of-the-art reproducible pipeline capable of representing simple to advanced analysis, reconstruction, and visualization workflows. For the first time, 3D datasets can be interactively visualized and interpreted as operations proceed. We will incorporate novel algorithms that can best leverage real-time analysis and integrate all tools into a generalized graphical user interface to be used across all fields utilizing 3D characterization, visualization, or analysis. This will address new demands for 3D visualization that accompanies the deluge of multidimensional data acquired across fields of research---including heat fluxes, rain / water distribution, chemical plume analysis, internal sealed machinery. In this project, we give particular focus to reconstruct the 3D structure of inorganic and biological materials from an angular range of 2D images collected by a scanning / transmission electron microscope (S/TEM). Volumetric / tomographic characterization requires advanced data analysis tools where the final 3D visualization is critically dependent on reconstruction algorithms that can take hours to days to complete. It would be extremely productive to see a reconstruction and begin analysis before it completes. The quality of a reconstruction will not only depend on the algorithm but also initial parameters---a bad reconstruction may cost scientists a day of waiting. Real-time visualization would allow researchers to make early judgments to best answer scientific questions. Here, for the first time, tomographic datasets can be interactively visualized and interpreted in 3D as operations proceed. We will incorporate novel algorithms that can best leverage real-time analysis and integrate all tools into a generalized graphical user interface to be used across all fields utilizing 3D characterization, visualization, or analysis. More specifically we will (1) establish real-time 3D analysis tool to capable of displaying live reconstructions (2) design an interactive 3D visualization platform that dynamically updates to incorporate new data as it arrives (3) develop real-time reconstruction methods for lower doses through compressed sensing. The project is open-source and can distributed between institutions.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jul 09, 2020
- Source ID
- W911NF2010147
Entities
People
- Robert Hovden
Organizations
- Army Contracting Command
- United States Army
- University of Michigan