A Monte Carlo model for energy spectra analysis in dedicated nuclear breast imaging

A Monte Carlo model for dedicated nuclear breast imaging was developed using the Monte Carlo N-Particle (MCNP) code. The modeled gamma camera was the LumaGem 3200s system, which comprised a 96 × 128 array of cadmium zinc telluride (CZT) elements with 1.6 × 1.6 mm² pixels and was equipped with a high sensitivity collimator. The patient model consisted of an 800-mL breast compressed to a thickness of 5.5 cm and an adjacent 8000-mL torso containing compartments modeling the liver and heart. Energy spectra from patients were acquired to determine an average patient energy spectrum. A phantom simulation was performed to determine the activity concentration in liver and heart regions versus the torso cavity and breast that produced an energy spectrum most closely matching the average patient spectrum. The Monte Carlo simulation was performed to simulate the energy spectra and breast images acquired from patient studies. The simulation allowed changes in intrinsic energy resolution of the detector and energy window. A correction to model the tailing effect in the CZT was also developed. The spectral components and their contribution to the energy windowed image were examined and the effect of changes in energy resolution on tumor contrast was determined. Results showed that 13-19% of counts in the breast image are scattered events (primarily first order Compton) and scatter from the torso region accounts for less than 4% of counts in the breast image for energy resolutions between 3.9-20%. Events from the torso are concentrated at the chest wall edge of the detector's field of view, consequentially decreasing tumor contrast in this area. Because of low overall scatter in the breast, changes in energy resolution were found to have minimal effect on tumor contrast.