Scatter/primary in mammography: Monte Carlo validation

The purpose of this investigation was to compare and validate the performance of the SIERRA Monte Carlo simulation routines for the analysis of the scatter to primary ratio (SPR) in the mammography setting. Two Monte Carlo simulation methods were addressed, the direct method was a straightforward and geometrically accurate simulation procedure, and the convolution method uses idealized geometry (monoenergetic, normally incident delta function input to the scattering medium) to produce scatter point spread functions (PSFs). The PSFs were weighted by the x-ray spectrum of interest and convolved with the held of view to estimate SPR values. The SPR results of both Monte Carlo procedures were extensively compared to five published sources, including Monte-Carlo-derived and physically measured SPR assessments. The direct method demonstrated an overall agreement with the literature of 3.7% accuracy (N = 5), and the convolution method demonstrated an average of 7.1% accuracy (N = 14). The comparisons were made over a range of parameters which included field of view, phantom thickness, x-ray energy, and phantom composition. Limitations of the beam stop method were also discussed. The results suggest that the SIERRA Monte Carlo routines produce accurate SPR calculations and may be useful for a more comprehensive study of scatter in mammography. (C) 2000 American Association of Physicists in Medicine. [S0094-2405(00)01808-3].