AI-Driven Scan-to-Discovery in 20 Minutes: Automated Segmentation of Synchrotron Micro-CT at Scale
This paper presents an automated AI-driven HPC pipeline that integrates distributed tomographic reconstruction with ensemble foundation models (SAM3 and DINOv3-Seg) to achieve near real-time, 20-minute scan-to-discovery segmentation of synchrotron micro-CT data, demonstrated through the analysis of grapevine xylem structures.