Analysis of the Use of Monoenergetic X-rays in Mammography

Abstract

This thesis studies the advantages of monoenergetic X-rays for mammography as a replacement for conventional broad-spectrum X-rays. Monoenergetic X-rays have the potential to enhance image quality with a reduced patient dose, although the extent of this potential is unknown. With the recent development of a practical monoenergetic X-ray source, where the X-rays are generated through X-ray fluorescence, it is prudent to examine their potential benefits. Monte Carlo Simulations were used in the study to produce images of breast phantoms containing signals (which mimic tumors) using a basic photon physics model. The phantom composition is 50% fibro-glandular/50% adipose tissue and varies in thickness from 3 - 10 cm. Ten signals of different thicknesses (from 1 to 10 mm) within a phantom are composed of 100% fibro-glandular tissue. Contrast-to-noise squared per mean glandular dose was established as the figure of merit (FOM) used to assess the performance of simulation output. We find that the optimal energy for our simulations varied with breast-phantom thickness and composition and that substantial performance improvements are observed compared to conventional spectral imaging. However, contrast-to-noise squared per mean glandular dose, the selected FOM, has known drawbacks as a performance indicator, with its deficiency in correlating actual signal detectability. Therefore, it is recommended that the study be continued and utilizes the industry standard ideal model observer, Hotelling detectability, as the FOM.