Independent benchmarking of five AI platforms for mandibular canal segmentation on CBCT: Ground-truth and AI-generated 3D STL dataset

This dataset contains de-identified 3D surface models (STL) of the mandibular canal used in the study “Independent benchmarking of five AI platforms for mandibular canal segmentation on CBCT images.” The dataset includes: Ground truth STL models of mandibular canals manually annotated by an expert radiologist. AI-generated STL models from five segmentation platforms: Relu Creator (Relu BV, Belgium), 3D Slicer (USA), Craniocatch (Türkiye), Atomica (USA), and Blue Sky Plan (USA). Important: No raw CBCT (DICOM) data are included. Only derived, anonymized STL meshes are shared.