OD130 - AGATA: advanced Geant4-based application for in-silico clinical trial in x-ray breast imaging

Purpose: New 3D breast x-ray imaging techniques, such as digital breast tomosynthesis (DBT) and breast dedicated CT (BCT), are being evaluated in specific clinical trials to study potential benefits in comparison to conventional 2D digital mammography (DM). These studies involve large patient cohorts for several years with the study ongoing in USA (TMIST) for the comparison of DBT and DM with a target enrollment of 160,000 patients. Virtual clinical trials represent an alternative solution to large scale human trials, with the prospect of reducing costs, time and issues related to the patient exposure to ionizing radiation. These are in-silico models of clinical trials with simulated devices and digital patient models. In this context, the INFN AGATA project aims at developing a platform for virtual clinical trials in breast x-ray imaging by means of digital models of patients obtained from high-resolution 3D breast images and a Monte Carlo simulation software. Materials and Methods: The digital breast phantoms were derived from 300 3D breast images acquired at UC Davis during clinical trials involving a BCT scanner. Each of the image voxels was classified as air, glandular tissue, adipose tissue or skin by means of a semi-automatic procedure. A mechanical model of the tissues permitted accurate compression to be applied in order to simulate DM or DBT examinations. The devices were simulated using Monte Carlo software based on the Geant4 toolkit. On the basis of the input exam specifications and modality, the software simulates the glandular dose distribution within the organ and the breast image projections. Results: The AGATA platform computed the dose distribution assuming ideal detectors for the image acquisition in DM, BCT or DBT, relying on digital breast phantoms. The developed software tracked up to 7×10 4 photons/s on a 16-core Dual-CPU AMD Opteron(tm) Processor4284. An ad-hoc cluster of servers reduces the computational times down to practical levels. Conclusions: The developed platform represents an alternative strategy for testing and developing devices in the field of 2D and 3D breast x-ray imaging. It permits access to patient-like data to a broad range of researchers and for comparing many technical solutions and imaging modalities by avoiding patient exposure to ionizing radiation