Stopping-power prediction with dual-energy computed tomography

Abstract

A substantial component of range uncertainty in ion therapy is attributed to the prediction of ion stopping-power ratio (SPR) from x-ray computed tomography (CT). Besides imaging technology that uses the same particles as are employed for treatment (e.g., proton CT), dual-energy CT (DECT) has been proposed to potentially improve this prediction within the framework of clinical x-ray CT. The aim of this thesis was to demonstrate improved SPR prediction with DECT from theoretical, experimental and clinical viewpoints. In the first step, an optimized approach was developed on the basis of a rigorous theoretical framework. Since variability in the SPR of human tissue is dominated by electron density, its universal and accurate determination solves the larger part of the problem and reduces the empirical component to the stopping number, which represents the second factor in SPR. By inferring the stopping number from DECT-derived photon cross sections, linear mixing behavior was demonstrated, which enabled proper calibration and the possibility to quantify uncertainty. The optimized approach was experimentally verified and benchmarked against the clinical gold standard in homogeneous animal tissues (human-like composition) and an anthropomorphic head phantom (human-like geometry). Furthermore, significant range differences to the standard approach were observed in patients, highlighting clinical relevance. From a methodological and experimental perspective, the developed method of SPR prediction with DECT outperforms the clinical gold standard and reduces the associated uncertainty to below 1%, which might eventually lead to the reduction of treatment margins. The thesis is presented in cumulative format and comprises eight peer-reviewed publications.