The influence of scattered photons on the accurate determination of microcalcification thickness in digital mammography

Abstract

Our interest has been to study the effect that scattered radiation has on contrast, signal-to-noise ratio and thickness reconstruction in digital mammographies. Using the GEANT code we have performed Monte-Carlo simulations of 25 kVp Mo/Mo photons, through a breast phantom which contains a 0.2-1.0 mm thick microcalcifications incident on a 20x106 $mm^{2}$ pixelized detector. The data have been analyzed assuming 6 different shapes of the incident beam: a 0.2x0.2 $mm^{2}$ ``narrow'' beam, 4 different 20 mm long scanning beams of various widths, and a 20x100 $mm^{2}$ beam with no scatter reduction mechanisms (NSR) . Since the image of a point depends on scattered photons which passed up to 2 cm away from the object (for 4 cm thick phantom), we identify the background definition as a main source of systematic uncertainty in the image quality analysis. We propose the use of two dimensional functions (a polynomial for the background and Gaussians for the signal) for total photon transmission description. Our main results indicate the possible calcification thickness reconstruction with an accuracy of the order of 6% using 3 mm wide scanning beam. Signal-to-noise ratio with the 3 mm wide beam gets improved by 20% with respect to NSR, a figure similar to that obtained with the narrow beam. Thickness reconstruction is shown to be an alternative to signal-to-noise ratio for microcalcification detection.