A dataset of 201 healthy elbow CT scans with humeroulnar joint segmentation, 3D reconstruction, and anatomical feature annotations

Accurate 3D representations of elbow joint anatomy derived from CT imaging are essential for total elbow arthroplasty (TEA) planning, biomechanical modelling, and patient-specific implant design. However, publicly available datasets that support such applications remain scarce. Here we present the first open, large-scale dataset that pairs raw elbow joint CT scans and segmented humeroulnar joint masks with expert-validated 3D bone models and quantitative morphometric measurements. The dataset includes de-identified CT scans from 201 individuals aged 16–86 years with no elbow pathology. Orthopaedic surgeons manually segmented the distal humerus and proximal ulna to generate STL-format 3D surface meshes. Ten clinically relevant anatomical parameters related to TEA were measured using standardized protocols. The dataset is expected to accelerate development of automated segmentation algorithms, validate finite element models, and support population-matched implant design. All data are publicly available under a CC BY 4.0 license via the Zenodo platform.