XCAT 3.0 Personalized Phantom Library

Description

Resource Type: Dataset

XCAT 3.0 is a comprehensive library of personalized computational phantoms designed to support virtual imaging trials, imaging system assessment, and AI development. The resource was created using automated large-scale organ segmentation and quality-control pipelines applied to clinical CT datasets, producing over 2,500 anatomically detailed virtual patients containing approximately 140 segmented structures. Phantoms are available in both voxelized and surface-mesh representations and can be integrated with physics-based imaging simulators such as DukeSim to generate realistic synthetic CT data. The library enables controlled studies of population variability, imaging protocol optimization, disease simulation, and virtual reader evaluation across diverse clinical and technical scenarios.

Reference: 10.1016/j.media.2025.103636

Website: https://xcat-3.github.io/