Simulation of Hounsfield Units for a Computed Tomography Scanner and Different Phantom Inserts

The calculation of carbon ion range in tissue for use in heavy ion treatment is based on the Hounsfield unit look-up table (HULT) [1]. Assuming that measured HU is correct, the HULT enables fast and reliable translation of HU to water equivalent depth (WED). However, HU values can vary with measurement parameters, such as X-ray voltage and the image reconstruction algorithm. The correlation between these parameters and HU is known only empirically. In this project, we present an approach to simulate HU that takes into consideration the geometry, composition, and physical process behind the measurement of HU values. The approach involves Monte Carlo simulations followed by a simple filtered back-projection reconstruction. The Monte Carlo code used is BEAMnrc/EGSnrc. Simulations of an X-ray tube, including the associated filters and beam shapers of a Siemens Emotion CT, were used to generate the initial beam shape and spectra. We performed a number of further simulations with various phantoms and substitutes. The resultant phase space files were analyzed to calculate the projections, taking into account the energy response of the CT-detectors. Then, a simple reconstruction algorithm (FBP using a Ram-Lak filter) was applied to the calculated projections and finally the data were presented in HU format. The above approach was developed for a Siemens Emotion CT with a DURA352 X-ray tube [2].