FleXray: Universal Anatomical Segmentation Model for Clinical X-rays

Forum topic · 小凯 · 2026-09-24

Summary

FleXray is a generalist deep learning model for anatomical segmentation across the entire body in clinical X-rays, developed by researchers including Victor Ion Butoi, Adrian V. Dalca, and Neel Dey (arXiv:2609.26756). X-ray remains medicine's most used yet least quantitative imaging modality because 3D anatomy is collapsed into 2D projections with overlapping structures and ambiguous boundaries, making manual labeling impractical. Rather than curating annotated datasets, the authors build a scalable, physics-based generative X-ray data engine that combines existing 3D whole-body CT segmentation datasets with generative image-editing models to simulate fully-annotated 2D X-rays with diverse appearances, physiological properties, and imaging geometries. Trained on these simulations, FleXray accurately segments 60 anatomical structures on unseen research datasets and in-the-wild X-rays. The model enables automated quantitative measurements for disease grading, robust navigation during X-ray-guided interventions, and data-efficient learning of pathological targets. Model, code, a full-body segmentation dataset, and a browser-based tool are released at https://flexray.csail.mit.edu.

FleXray: Universal Clinical X-ray Segmentation

Field: Computer Vision Authors: Victor Ion Butoi, Vivek Gopalakrishnan, John V. Guttag, Adrian V. Dalca, Neel Dey Posted: 2026-09-22 arXiv: 2609.26756

Abstract

X-ray is medicine's most widely used imaging modality, yet remains among its least quantitative. Unlike volumetric modalities like CT or MRI, X-ray collapses 3D anatomy into a 2D projection, causing structures to overlap and anatomical boundaries to be ambiguous, even to experts. As a result, labeling X-ray databases for training general-purpose segmentation systems is impractical, leaving morphometric and functional X-ray analysis confined to narrow anatomical regions and applications.

To this end, we present FleXray, a generalist model for anatomical segmentation across the entire body in clinical X-rays. Instead of curating large, manually annotated X-ray datasets, we build a scalable, physics-based generative X-ray data engine. Using existing 3D whole-body CT segmentation datasets and generative image-editing models, we simulate fully-annotated 2D X-rays with diverse appearances, physiological properties, and imaging geometries.

Trained on these simulations, FleXray accurately segments 60 anatomical structures across unseen research datasets and in-the-wild X-rays. We further show that FleXray makes X-rays directly amenable to quantitative analysis, enabling automated measurements for disease grading, robust navigation during X-ray-guided interventions, and data-efficient learning of pathological targets.

We release the model, code, a full-body X-ray segmentation dataset, and a local, easy-to-use browser-based tool at https://flexray.csail.mit.edu .

Key Points

Problem: X-rays lack quantitative segmentation tools because 2D projection creates overlapping structures and ambiguous boundaries, making manual annotation impractical at scale.
Approach: A physics-based generative X-ray data engine that leverages existing 3D whole-body CT segmentation datasets plus generative image-editing models to synthesize fully-annotated 2D X-rays.
Capability: Segments 60 anatomical structures across unseen research datasets and real-world X-rays.
Applications: Automated measurements for disease grading, navigation for X-ray-guided interventions, and data-efficient learning of pathological targets.
Release: Open model, code, dataset, and local browser-based tool at https://flexray.csail.mit.edu .