2005 International Nuclear Atlantic Conference - INAC 2005 Santos, SP, Brazil, August 28 to September 2, 2005 ASSOCIAÇÃO BRASILEIRA DE ENERGIA NUCLEAR - ABEN ISBN: 85-99141-01-5 MONTE CARLO SIMULATION OF X-RAY SPECTRA IN DIAGNOSTIC RADIOLOGY AND MAMMOGRAPHY USING GEANT4 Daniel A. B. Bonifácio 1 , Hélio M. Murata 2 and Maurício Moralles 3 1 Departamento de Física Geral - Instituto de Física Universidade de São Paulo Rua do Matão, Travessa R, 187 05508-900 São Paulo, SP daniel@if.usp.br 2 Instituto de Eletrotécnica e Energia Universidade de São Paulo Av. Professor Luciano Gualberto, 1289 05508-900 São Paulo, SP murata@iee.usp.br 3 Instituto de Pesquisas Energéticas e Nucleares (IPEN / CNEN - SP) Av. Professor Lineu Prestes 2242 05508-000 São Paulo, SP moralles@ipen.br ABSTRACT The open-source object-oriented toolkit GEANT4 was used to simulate x-ray spectra in diagnostic radiology and mammography. The simulations were performed using different combinations of target, filters and tube voltages. All the relevant physical processes were included in the calculations: Compton scattering, photoelectric effect, Rayleigh scattering, bremsstrahlung and ionization. The analyzed energy range is from 10 keV to 150 keV. Both Penelope and Low Energy physical models included in the Low Energy extensions of GEANT4 toolkit were used in this work. Range cuts for electron and gamma were set to 500 nm and 3000 nm, respectively. The simulated x-ray spectra using both physics models were compared with calculated spectra generated by the IPEM report number 78. Results show good agreement for the bremsstrahlung intensity for the spectra with tube voltages 40 kV, 100 kV and 150 kV, while the bremsstrahlung intensity is larger for the simulated spectra with 25 kV and 30 kV. Simulated characteristic peaks present lower intensities all spectra. These discrepancies should be related with the ionization process and/or the atomic relaxation implemented in the code. The cross section tables for electrons used in the simulations should be checked. 1. INTRODUCTION One of the most important factors to protect the patient from radiation and to optimize medical diagnostic radiology is to understand the relationship between radiation dose and image quality. This aim can be reached with the knowledge of the diagnostic x-ray energy spectra, which provide a complete description of the x-ray beam. Since Kramers’ first attempt in 1923, several research groups are working to find an accurate method for predicting x-ray spectra, which would be very useful because experimental measurement of x-ray spectra[1,2] is time consuming and requires special equipment which is available only in some laboratories. There are three categories of methods for x-ray spectra prediction: empirical models[3], semi- empirical models[4,5] and Monte Carlo[6,7,8] simulations. The main advantage of empirical and semi-empirical models is the low computation time consumption. Otherwise these