发表机构
Instituto de Física Corpuscular (CSIC-University of Valencia); Universitat Politècnica de Catalunya; GSI Helmholtzzentrum für Schwerionenforschung; Institut für Kernphysik, Technische Universität Darmstadt; West German Proton Therapy Centre（粒子物理研究所（西班牙科学理事会-瓦伦西亚大学）; 加泰罗尼亚理工大学; 德国重离子研究中心; 达姆施塔特工业大学核物理研究所; 西德质子治疗中心）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本研究在等时回旋加速器上实验验证了混合PGI-PET三维成像用于质子射程验证，PET精度达0.8毫米，展示了其临床应用的互补优势。
AI 中文摘要
射程不确定性仍然是质子治疗全面临床开发的主要限制，这促使了可靠的体内射程验证技术的需求。本研究在德国西部质子治疗中心的等时回旋加速器上，在临床代表性的笔束扫描条件下，对混合瞬发伽马成像（PGI）—正电子发射断层扫描（PET）概念进行了实验研究。由于该加速器提供准连续束流，因此采用了脉冲式“射击-步进”结构来模拟与临床能量层切换相关的中断。这使得在同一序列中，在辐照期间进行PGI，在束流关闭间隔内进行PET和延迟康普顿成像成为可能。四台双平面康普顿相机以共面十字形几何结构围绕束流等中心布置。用100 MeV质子沿束流轴在已知位置照射聚乙烯体模。通过监督机器学习模型从重建的一维剖面中估计其位置，这些模型仅使用蒙特卡洛模拟进行训练。PET取得了最佳性能，均方根偏差为0.8毫米，其次是PGI，为1.6毫米。基于e$^{+}$湮灭光子和$^{10}$C$^{*}$衰变的延迟康普顿成像分别产生了2.4毫米和3.8毫米的结果。PET在事件统计量降至原始水平的1%时仍保持良好的性能，而PGI则逐渐受到统计限制。这些结果证明了混合PGI-PET的互补优势，并支持其在临床相关辐照条件下向毫米级质子射程验证方向发展。
英文摘要
Range uncertainties remain a major limitation to the full clinical exploitation of proton therapy, motivating reliable in-vivo range-verification techniques. This work presents an experimental investigation of a hybrid Prompt-Gamma Imaging (PGI)--Positron-Emission Tomography (PET) concept at the isochronous cyclotron of the West German Proton Therapy Centre under clinically representative pencil-beam-scanning conditions. Because this accelerator delivers a quasi-continuous beam, a pulsed shoot-and-step structure was imposed to mimic the interruptions associated with clinical energy-layer switching. This enabled PGI during irradiation and PET and delayed Compton imaging during beam-off intervals within the same sequence. Four two-plane Compton cameras were arranged in a co-planar cross-shaped geometry around the beam isocenter. Polyethylene phantoms were irradiated with 100~MeV protons at known positions along the beam axis. Their positions were estimated from reconstructed one-dimensional profiles using supervised machine-learning models trained exclusively on Monte Carlo simulations. PET achieved the best performance, with a root-mean-square deviation of 0.8~mm, followed by PGI with 1.6~mm. Delayed Compton imaging based on e$^{+}$ annihilation photons and $^{10}$C$^{*}$ decays yielded 2.4~mm and 3.8~mm, respectively. PET retained good performance down to 1% of the original event statistics, whereas PGI became increasingly statistics-limited. These results demonstrate the complementary strengths of hybrid PGI--PET and support its development towards millimetre-scale proton-range verification under clinically relevant irradiation conditions.
