Abstract
BACKGROUND: In MRI-integrated radiotherapy, image registration between magnetic resonance imaging (MRI) and computed tomography (CT) can introduce systematic errors of up to 5 mm. To avoid such errors in proton therapy, one of the prerequisites is the determination of the tissue-water linear stopping power ratio Stissue, water from MRI for treatment planning. PURPOSE: Stissue, water depends on the elemental composition of the medium. Proton density-weighted MRI measures the concentration of hydrogen ( 11H ), which is independent of the magnetic field strength. This study evaluated the potential of proton density-weighted MRI to estimate Stissue, water in MRI-integrated proton therapy. METHODS: Based on ICRU 46, we analyzed and modeled the relationship between tissue-water hydrogen concentration ratio Htissue, water and Stissue, water at 100 MeV proton energy using linear regression. The model was evaluated for the accuracy of the regression fit and tissue composition variability at proton energies from 70-230 MeV. To assess the accuracy of proton density-weighted MRI, we developed a deuterium oxide (D(2)O)-water (H(2)O) phantom that replicates the hydrogen concentration range in human tissues. Fast spin-echo (FSE) and gradient-echo (GRE)-based sequences were compared. Positional and voxel-level signal variability were investigated. RESULTS AND DISCUSSION: For soft and bone tissues, there was a linear correlation (R(2) = 0.99) between Htissue, water and Stissue, water . The inflated lung deviated from this correlation because it includes air volume, which reduces Htissue, water and Stissue, water significantly. By incorporating air and compressed lung from ICRP 110, a linear correlation (R(2) = 1.00) was found for lung-related tissues. The D(2)O-H(2)O phantom covered the hydrogen concentration range ( Hsolution, water = 0.30-1.00) relevant to human tissue. A partial molar volume effect in the phantom emphasized the need for mass density measurement. The FSE sequence provided higher image quality and demonstrated a strong linear correlation (R(2 )= 1.00) between Hsolution, water and signal-noise ratio (SNR). By contrast, the GRE showed distortion artifacts. Based on the evaluation criteria, composite uncertainties were 6.89%, 3.00%, and 1.92% for soft, bone, and lung tissues, respectively. Adipose tissue contributed significantly to soft tissue uncertainties. CONCLUSIONS: The relationship between Htissue, water and Stissue, water remained consistent despite variability in tissue composition and treatment energies. The D(2)O-H(2)O phantom, which is simple and reproducible, proved effective for accurately calibrating proton density-weighted MRI against Htissue, water . These findings demonstrate the potential of proton density-weighted MRI to directly estimate Stissue, water . A separate method to identify adipose through lipid concentration measurement may further improve accuracy in soft tissues.