LAUSR

Accurate determination of the stopping power ratio (SPR) of the tissues along the beam path is required for calculations of dose deposition in proton therapy. Schneider et al. [1] proposed a stoichiometric calibration method to convert computed tomography (CT) numbers to SPRs via a fitted calibration curve, which is commonly used [2]. However, this conversion introduces an uncertainty of 3–3.5% in proton range [2–6]. To minimize the impact of range uncertainty on treatment plans, either larger margins distal and proximal to the tumor are added or robust optimization must be performed [7,8]. This practice may result in irradiating surrounding healthy tissues more than necessary. Dual-source dual energy CT (DECT) was proposed to address such challenges [6,9]. SPR can be calculated on the basis of the Bethe formula utilizing the relative electron density (qe) and mean excitation energy (Im) of tissues. Yang et al. [6] first proposed to use qe and the effective atomic number (Zeff) obtained from DECT to calculate SPR with a linear fit between Zeff and ln Im. Bourque et al. [10] later proposed a continuous association between Zeff and Im, which was well adopted in subsequent studies to estimate Im from Zeff [3,11]. Hudobivnik et al. [12] demonstrated a higher accuracy in SPR derived from DECT than from single-energy CT (SECT). The accuracy of DECT approach was validated in a ground-truth anthropomorphic phantom [13], and its clinical relevance was demonstrated in patient-cohort analyses [14]. The root-mean-square error (RMSE) of SPR with a DECT approach was 1%, in comparison with proton beam measurements [15,16]. Dual-layer CT (DLCT) is a newer type of DECT with a novel detector design [17]. DLCT splits the polychromatic X-ray spectrum of a single CT scan into lowand high-energy spectra on a detector level to generate qe, Zeff, and virtual monoenergetic (MonoE, 40–200 keV) images. A recent evaluation of the first commercial DLCT system reported that the accuracy of qe was better than 1%, as compared with that of reference values, and the deviation in Zeff was within ±2%, except for lung tissues [18]. Studies on DLCT for particle therapy applications are scarce. Feller et al. [19] reported 0.6% accuracy of SPR prediction for tissue-mimicking materials (TMMs) by comparing calculated and measured SPRs with carbon ion beams. Landry et al. [20] showed that <1% RMSE in SPR can be achieved with DLCT for TMMs. Nevertheless, lung TMM inserts exhibited larger errors for all quantities in their study and were therefore excluded from the reported RMSE values. The chemical compositions of fresh tissues differ from those of TMMs, which can affect the proton beam transport. The experimental validation of proton range with DLCT using real tissues is pertinent and much needed. In this study, we determined the accuracy of SPR estimated from DLCT and performed the first experimental validation of proton ranges calculated in a treatment-planning system (TPS) for both TMMs and animal tissues.