Design and measurement of 3D printed variable-morphology and variable-density breast phantoms for mammography and breast CT dose assessment

Accurate dose quantification across radiation-based breast imaging techniques is important for patient safety, but currently available phantoms do not easily allow the assessment of the impact of breast size and density across modalities. Here we design, produce and validate 3D printed, variable-morphology, fillable phantoms for the purpose of radiation dose measurements and comparison studies. Three representative breast shapes (small, medium, large) were exported from real BCT data. Corresponding compressed shapes were generated following the Virtual Imaging Clinical Trials for Regulatory Evaluation (VICTRE) pipeline, in which morphology is controlled through 10 separate length and deformation parameters, optimized to match uncompressed phantom volumes. Surface models of each shape were printed using a 3D printer (Stratasys F370-3D, alphacam GmbH). Four mixtures, representative of BI-RADS categories A-D, were prepared using varying ratios of 1.5% agar, rapeseed oil, and soy lecithin, poured into the printed surface models, and set overnight at 4°C. Pre-calibrated MOSFET dosimeters were placed on the irradiated surface of each phantom to measure Entrance Surface Dose (ESD). Compressed phantoms in all sizes and densities were successfully measured in a mammography device (Senographe Essential, GE Healthcare, DE). Uncompressed phantoms in all sizes and in three of the four densities were successfully measured in a BCT device (nu:view, Advanced Breast-CT). Measured ESD matched the values obtained with Monte Carlo simulations; for the unit equipped with automatic exposure control, the ESD increased with increasing breast size and with increasing breast density, confirming the known correlation between thickness, density and ESD and validating the capabilities of the proposed phantoms to represent breasts with real physical properties.