CNR performance of semiconductor materials for X-ray imaging of breast calcifications

Abstract

Abstract We present the results of a GAMOS/GEANT4 computer simulation of a standard X-ray mammography system, which consists of a Tungsten 28 kVp polychromatic X-ray source with a 50 μm Rh filter, a mammography phantom with Al2O3 spherical specks of different diameters, and a generic pixel detector (55 μm × 55 μm pixel size) with different types of semiconductor sensors. The number of photons simulated is calibrated to produce similar entrance surface dose (ESD) as the one used by a standard clinical mammography screening. Estimates of Contrast to Noise Ratio (CNR) as a function of ESD, sensor thickness and microcalcification diameter are presented for four different sensor materials: Silicon (Si), Cadmium Telluride (CdTe), Gallium Arsenide (GaAs) and Perovskite (MAPbI3). For the X-ray energy spectrum and pixel size considered, and an ESD dose of 4 mGy, our study shows that, with the exception of Si, these sensors, as thin as 200 μm, are able to resolve (with at least 3 standard deviations above background) Al2O3 spherical specks up to a minimum diameter of 180 μm, having statistically compatible CNR performance. The increase in substrate thickness has a substantial improvement in the CNR values provided by the Si sensor, while for the other cases the enhancement of CNR is marginal and consistent with statistical uncertainties with the thinnest case considered.