Implementation and 3D extension of dose reconstruction strategies from prompt-gamma emissions in proton therapy

<i>Objective.</i>Range and dose monitoring with secondary radiation can help minimizing the issues of range uncertainties in proton cancer therapy. Prompt gammas (PGs) have been widely investigated as a promising secondary radiation for<i>in vivo</i>verification. Since it can be argued that for a proper delivery of the intended treatment the most desirable quantity to assess is the dose distribution<i>in vivo</i>, this work aims at the reconstruction of the delivered proton dose from distributions of PG radiation.<i>Approach.</i>Some techniques have already been proposed in the literature to reconstruct the dose from a distribution of detected secondary radiation, mostly positron emitters. Among them, very promising methods are the analytical deconvolution approach, the evolutionary algorithm and the maximum-likelihood expectation-maximization (MLEM) algorithm. Herein, the feasibility of the application of these approaches to PG distributions at emission stage is assessed with simulated mono- and polyenergetic proton beams, irradiating homogeneous and inhomogeneous phantoms, and a realistic case of a head and neck (H&N) tumor patient.<i>Main results.</i>The accuracy of the reconstructed dose is evaluated via comparison with the corresponding simulated ground truth dose distributions using different metrics. For the case of 1D reconstruction on phantoms, theΔR80,ΔR50andΔR10, withΔR%being the difference of the positions at the % of the dose maximum in the distal fall-off region between the simulated and reconstructed curves, are always less or of the order of 1 mm in absolute value. For 3D reconstruction on phantoms and on the H&N case, the<i>γ</i>(1%/1 mm) passing rate is always above or equal to 97%.<i>Significance.</i>This study demonstrates the applicability of the analytical deconvolution, the evolutionary and the MLEM algorithms to dose reconstruction from PG emissions, providing a step forward toward the final goal of real-time verification of the dose delivery for real-time adaptive particle therapy.