The density measurements in pCT imaging

Abstract

In existing proton treatment centers, dose calculations are performed based on x-ray computerized tomography (CT). Alternatively, the therapeutic proton beam could be used to collect the data for treatment planning via proton CT (pCT). With the development of medical proton gantries, first at Loma Linda University Medical Center and now in several other proton treatment centers, it is of interest to continue the early pCT investigations of the 1970s and the early 1980s. From that time, the basic idea of the pCT method has advanced from average energy loss measurements to an individual proton tracking technique. This reduces the image degradation due to multiple Coulomb scattering. Thereby, the central pCT problem shifts to the fidelity of the physical information obtained about the scanned patient, which will be used for proton treatment planning. The accuracy of relative electron density distributions extracted from pCT images was investigated in this work using continuous slowing down approximation (CSDA) and water-equivalent-thickness (WET) concepts. Analytical results were checked against Monte Carlo simulations, which were obtained with SRIM2003 and GEANT4 Monte Carlo software packages. The range of applications and the sources of absolute errors are discussed.

Comparison of GEANT4 Simulations with Experimental Data for Thick Al Absorbers

“…The CSDA values were calculated based on the same form of Bethe-Bloch equation as it had been published in [10,11]. The calculations of proton energy spectra using Payne´s theory [18] were based at the equations which could be found in [10,11] also.…”

“…Its ability to work with very complex geometry and tracking is essential to the pCT development [10][11][12][13][14]. The simulations were made with the version GEANT4.8.2 by adjusting the parameters for the Hadrontherapy toolkit, as it is discussed in [19].…”

“…At least, our previous simulations [10][11][12][13][14] have shown that there are notable differences between the spectra, simulated by GEANT4 (in the standard execution mode) and the results of simulation by TRIM/SRIM2003 [15] and MCNPX [16]. These differences in the mean final proton energy are observed for H 2 O and Al absorbers at any thickness, but for relatively thin (less than ~50% of the full range of protons with the same energy in the same material) targets are less than it could be resolved in a real experiment.…”

“…In addition, it has been observed that there is an obvious scale effect: all the spectra simulated for the pairs of initial proton energies 25MeV -250MeV [11][12][13], and 23MeV -230MeV [14], could be scaled one to the other by applying a factor 10 to the energy scale. It should be noted that the spectra were simulated for target thicknesses equal to 10% to 90% of the corresponded full projected range from NIST PSTAR.…”

GEANT4 Tuning For pCT Development

“…The pCT deals specifically with relatively thick targets, like the human head or trunk. Our initial comparative simulations [4], [5], [6] with GEANT4 and TRIM from the SRIM2008 package [7], [8] have shown that these two popular Monte Carlo codes usually generate somewhat different proton spectra. In the case of thin absorbers, the difference could be ignored because it is less than could be detected experimentally.…”