Investigating the dependence of breast calcifications signal to noise ratio on mammographic spectra: Monte Carlo simulation studies

Abstract

A Monte Carlo simulation model is utilized to study the influence of mammographic spectra on the SNR of simulated inhomogeneities representing calcifications of various thicknesses, embedded inside a mathematical breast phantom. The phantom is semicylindrical, mimicking a physical phantom for mammography, with background simulating glandular breast tissue. A step wedge arrangement with foils composed of Hydroxyapatite, was embedded. Monte Carlo methods have been used to sample the initial x-ray photon energy and trace the photons inside the breast, taking into account all possible physical processes. For the purposes of this study, only the third dimension (thickness) of calcifications was kept in realistic size, since it is the one related with the influence of the x-ray spectrum on the image, while the calcifications visibility with respect to their projection size is mainly affected by the resolving power of the imaging system. The discrimination of different calcifications thicknesses, described in terms of SNR, strongly depends on all the parameters affecting the x-ray spectrum. Tube voltage strongly affects SNR of small and medium thickness calcifications. The common Mo/Mo spectrum demonstrates improved characteristics, with SNR differences up to 14%, with respect to Rh/Rh for the entire thickness range of calcifications studied. Increase in filter thickness decreases SNR for calcification thicknesses less than 1200 µm, while beyond this point increased filter thickness results in improved SNR. As far as the filter material is concerned, the thicker calcification to be visualized, the higher the k-absorption edge of the filter required, in order to achieve the maximum SNR.