Patient-derived compressed digital breast phantom dataset for mammography and digital breast tomosynthesis simulations

The dataset contains 821 digital breast phantoms derived from patient dedicated breast CT (bCT) scans and mechanically compressed in the cranio-caudal (CC) view using a validated finite-element compression model. Each phantom originates from a segmented bCT volume (isotropic voxel size of 0.273 mm), where voxels were classified into: 0 – air, 1 – adipose tissue, 2 – fibroglandular tissue, and 3 – skin. The phantoms are provided as NRRD volumes at the original scan resolution.

The dataset is intended to support simulation studies in mammography and digital breast tomosynthesis, with applications including system optimization, radiation dosimetry, image reconstruction, and the development and validation of AI-based methods. The corresponding uncompressed phantoms are also included.