Directional scatter imaging as a confirmation tool for the presence of tumoral masses and calcifications in dual-projection mammography: Proof of concept study

Abstract

The purpose of this work is to determine, via Monte Carlo simulations, if the x-ray radiation scattered by an irradiated object at the energies typically used in mammography, contains imaging information that could be used to potentially confirm the presence of tumoral masses and calcifications without the need to further expose the patient. An anthropomorphic breast phantom, with a resolution of 100 μm, was used to simulate mammographic images. The anthropomorphic breast phantom, was compressed using the biomechanical software FEBio to obtain a breast model with two different thicknesses. Tumoral masses and microcalcifications of varying sizes were digitally added. Two imaging detectors placed at 90° from each other, one for the primary and the second for the scatter image, were explicitly modeled as amorphous selenium screens. The pixel size was assumed to be 74 μm and both the absorbed dose imparted as well as the number of scatter photons reaching the detector were calculated in a pixel matrix consisting of 3072 × 3072 elements. Two different mammographic x-ray spectra were used, with peak energies of 28 keV and 35 keV. Image noise was added by randomly modifying the calculated absorbed dose in a given pixel assuming the modification can be sampled from a Box-Muller approximation of a Gaussian distribution. The irradiation of the phantom is assumed to proceed using a fan beam sequentially delivered on a slice-by-slice basis so that a directional scatter image from each irradiated phantom segment can be independently obtained. A simple strategy to increase the scatter signal reaching the detector by adding the signal from adjacent phantom slices is shown to be able to reduce the imparted dose by a factor proportional to the number of slices used to form the scatter image. We have shown that despite the inherent high noise present, directional scatter images are still able to provide information that may be useful in corroborating whether or not a tumoral mass discernible in the primary image is an actual structure of interest or an image artifact.