Development of a Realistic Computational Breast Phantom for Dosimetric Simulations

*Currently, x-ray breast imaging technique is the gold standard in both screening and diagnosis of breast cancer. However, x-rays, as a form of ionizing radiation, also post a risk of inducing cancer in the patients. As screening is applied to a large proportion of the population, the radiation dose from breast cancer screening is carefully monitored and it should necessarily be so. Since the breast cancer often grows in the glandular tissues of the breast, the mean glandular dose (MGD) is the commonly used quantity to indicate the radiation risk. In the estimation of the MGD, the breast tissues are actually assumed to be a homogeneous tissue in the Monte Carlo simulations of the energy deposition in the breast; together with the measurement of the entrance skin exposure, the simulated energy deposition provides the MGD estimate to a patient. In reality, the glandular tissue does not distribute in the breast uniformly; the accuracy of such estimates with homogeneous breast tissues has been questioned. To address the issue, we have developed realistic computational breast phantoms with anatomical details and mammographic texture. The anatomical details are composed of geometric objects while the mammographic texture is generated from the Fourier transform of Gaussian noise. Our new method will have implications in the reconstruction of the MGD for individual patients from their mammographic examinations. Our new phantoms will also have applications in dosimetric and imaging studies across x-ray based breast imaging modalities. Another application of the new phantom will be in the assessment and characterization of the 3D image reconstruction – The new phantom will allow us to quantify the reconstruction error. In this work, we will use the phantom to validate the conventional assumption of homogeneous breast tissue in the MGD calculations. The phantom generation software will be made available to the community through an open source arrangement.