Healthy lung and heart reference volumes and dosimetry phantoms from automatic CT segmentation

This work aimed to establish average lung and heart volumes in a large population of healthy young Korean adults and to produce age- and sex-specific thoracic surface phantoms that precisely match those averages. The dataset is intended for researchers and clinicians in medical physics, radiology, and anatomy who require population-specific reference values and realistic three-dimensional models for simulation, device development, and education. Chest computed tomography scans from 2061 individuals (1361 males and 700 females in their 20s and 30s) obtained during routine health screenings were analyzed. Lung structures were segmented in all subjects and heart structures in 367 in 3D Slicer using TotalSegmentator extension. Volumes of seven pulmonary and nine cardiac structures were measured using Python. Multiple linear regression was used to summarize how volumes varied with age, height, and weight and to check internal consistency. For each sex and age group, a representative individual whose lung and heart volumes were closest to the group averages was selected. In Autodesk Maya, the lung and heart models from these individuals were uniformly scaled until the total lung and heart volumes deviated by less than 1% and 0.1%, respectively, from the group means resulting in four thoracic phantom models. The dataset includes four thoracic phantoms (20s male, 20s female, 30s male, 30s female), each containing 52 segmented structures provided as STL files (approximately 2 GB in total). Accompanying Excel files provide summary statistics and verification of scaled versus target volumes for all lung and heart structures. All files are available via Zenodo (https://doi.org/10.5281/zenodo.17388653) and the authors' homepage. These population-specific volume data and thoracic phantoms can be used as reference models for computational dosimetry, exposure and SAR assessment, medical device design and size matching, and anatomical or procedural education.