From Reconstruction to Interpretation: Zero-Setup Multi-Phase Segmentation of X-ray Tomography Data

The paper presents a zero-setup framework for multi-phase segmentation of synchrotron X-ray tomography data. It combines a material-agnostic mask preparation strategy with a pretrained semantic segmentation network and requires no user prompting, dataset-specific annotation, or deployment-time retraining. The system produces interpretable masks for background, sample, bright, dark-gray, light-gray, and porosity regions within minutes after reconstruction. The authors report evaluation on held-out synchrotron micro-CT images and qualitative tests on additional datasets, with physically meaningful results across samples and imaging conditions. The framework is positioned as a bridge from rapid reconstruction to near-real-time beamline interpretation.