DukeSeg Automated CT Segmentation Framework

Description

Resource Type: Software

DukeSeg is an automated CT segmentation framework developed to support virtual imaging trials, quantitative imaging analysis, and AI-based virtual reader research. Built upon modern deep learning segmentation architectures and trained using combined public and private datasets, DukeSeg generates detailed CT segmentations spanning approximately 140 anatomical structures. The framework was designed to enable scalable extraction of anatomical priors, body composition measurements, and patient-specific computational phantoms, including the generation of the XCAT 3.0 virtual patient library.