Multivariate Volume Visualization and Machine-Guided Exploration in Tomviz

Nanoscale imaging techniques employed at DOE’s modern X-ray facilities are important in guiding future experiments, enabling successful scientific discoveries, and bringing products to market. However, the increasing size and complexity of nanoscale imaging data have made the analysis process more difficult and, in some situations, infeasible. Nanoscale imaging techniques commonly produce multivariate volumetric data, which contains more than one channel per voxel and has become increasingly prevalent in important industries such as battery technology, manufacturing, and medical imaging. Due to its multidimensional nature, multivariate data is complex and usually requires substantial human effort to analyze. Often, each channel is visualized separately and individually. Such an approach, however, does not readily reveal the complex yet frequently important cross-channel interactions.We seek to develop a set of generic multivariate volume visualization and machine-guided analysis tools to be utilized within Tomviz for nanoscale imaging, as well as reused in additional applications for other important scientific domains such as medical imaging.We successfully explored several novel multivariate volume visualization and machine-guided analysis techniques through our set of proposed prototypes. Those techniques included informative visualizations, interactive machine-guided exploration techniques, and advanced multivariate segmentation and clustering techniques. We propose extending several of our core software systems in order to deliver production-ready bespoke multivariate volume visualization techniques into various computational environments and scientific domains.We build upon the invaluable discoveries made during the Phase I effort by evolving them into full commercial capabilities. Our Phase II work starts with VTK, ParaView, and trame for generalized multivariate volume analysis and visualization. It then integrates these techniques into Tomviz, modernizes Tomviz with web capabilities, and develops novel multivariate frameworks for empowering scientists through AI-driven annotations, data exploration, and segmentation. This work addresses the deficiencies in multivariate data visualization and analysis to grow a substantial share of the visualization and analysis market. In particular, we are targeting the battery, manufacturing and engineering, and medical imaging firms, in general, that can benefit from the underlying technology to reduce costs, improve productivity, and create innovative products.