Effect of X-ray scatter correction on the estimation of attenuation coefficient in mammography: a simulation study

A mammographic image requires high contrast for soft tissue imaging. Even small amounts of dispersion reduce the high contrast required to make accurate diagnoses. Current systems for digital mammography use an anti-scatter grid to reduce the scatter phenomenon. However, despite the widespread use of the anti-scatter grids in the clinical practice, it leads to elimination of useful primary radiation, thus forcing to the increase patient irradiation in order to achieve high contrast images. It is therefore desirable to develop digital image processing methods for scatter correction. The objective of this study is to evaluate how effective digital scatter removal can be achieved by implementing and tuning appropriate deconvolution functions by means of a simulation approach carried out on rectangular breast phantoms, with the ultimate aim of proposing a framework for the evaluation and comparison between experimental and theoretical breast attenuation coefficient as an indirect measure of the scattering effects in digital mammography. The phantom is composed of two types of step blocks representing the adipose and glandular tissue of the breast, provided by the manufacturer. In this study, it is assumed that the measured digital image is the result of the convolution between the primary image (devoid of scattering) and a spatially variant Point Spread Function, which represents the scattered radiation. The deconvolution of the measured image allows the recovery of the primary image and the assessment of the impact of scattering phenomenon on the attenuation coefficients of the examined sample.