AI-enabled processing of multimodal data to expand 2D chemical information to 3D space
EMSL Project ID
60674
Abstract
This project will develop a user-friendly workflow to process EMSL’s multi-modal data using AI segmentation algorithms in Fiji software (already available on EMSL OnDemand). This multi-modal data includes density-based images (grey-scale) from X-ray Computed Tomography (XCT) of soil samples and 2D chemical data from electron microscopy/spectroscopy. EMSL lacks the capability and workflow to merge the multi-modal data streams for a comprehensive understanding of soil organic matter distribution at pore space coupled with soil mineral chemistry in 3D space. Our research is timely and impactful because (1) EMSL users showed great interest in the 3D characterization of soil biogeochemistry advertised during exploratory proposals, (2) it benefits the current EMSL user program (e.g., Devin Rippner from USDA; #ProjectID: 51847; https://search.emsl.pnnl.gov/?project[0]=projects_51847) to improve AI/ML method for segmentation, and (3) is crucial to advancing multi-scale models (e.g., pore-scale to continuum-scale) from the data collected by MONet soil research platform. The project deliverable is a workflow that will be available to users from EMSL OnDemand. A successful outcome will be making this workflow user-friendly so that experts and non-experts can use it on their respective datasets when advertising large-scale research (LSR) proposals and MONet solicitations. Impact – Our project will provide data analytics tools to EMSL users to incorporate different components of our AI workflow into their own AI automation pipelines through Tahoma. Our reproducible workflow will be associated with demos that describes the EMSL capability will benefit its users and the BER community. Our digital products will be available to EMSL/BER community through EMSL OnDemand.
Project Details
Start Date
2023-01-30
End Date
N/A
Status
Active
Released Data Link
Team
Principal Investigator
Co-Investigator(s)
Team Members