Accuracy of the Most Likely Path Formalism in Inhomogeneous Phantoms

Proton computed tomography (pCT) has become a lively research field in medical imaging. Its importance lies in its ability to accurately locate the Bragg peak where the tumor is positioned for proton therapy treatment planning. The quality of the pCT image is primarily affected by the spatial resolution and relative electron density resolution. A measure of the spatial resolution is the amount of expected deviation of the actual proton paths from the theoretically derived paths based on the experimentally available data, the so-called most likely paths (MLPs). The MLPs are derived using the assumption that the object to be imaged is homogeneous water. Geant4 Monte Carlo simulations were used to simulate the actual proton paths through some inhomogeneous phantoms and were compared with MLP calculations. Statistical analyses were conducted to determine the spatial resolution of the protons in different phantoms as a function of inhomogeneity location, amount, and density.