Simard Mikaël, Fullarton Ryan, Volz Lennart, Schuy Christoph, Chung Savanna, Baker Colin, Graeff Christian, Fekete Charles-Antoine Collins
Department of Medical Physics and Biomedical Engineering, University College London, London, UK.
Biophysics, GSI Helmholtz Centre for Heavy Ion Research GmbH, Darmstadt, Germany.
Med Phys. 2025 Sep;52(9):e18081. doi: 10.1002/mp.18081.
Integrated mode proton imaging is a clinically accessible method for proton radiographs (pRads), but its spatial resolution is limited by multiple Coulomb scattering (MCS). As the amplitude of MCS decreases with increasing particle charge, heavier ions such as carbon ions produce radiographs with better resolution (cRads). Improving image resolution of pRads may thus be achieved by transferring individual proton pencil beam images to the equivalent carbon ion data using a trained image translation network. The approach can be interpreted as applying a data-driven deconvolution operation with a spatially variant point spread function.
Propose a deep learning framework based on paired proton-carbon data to increase the resolution of integrated mode pRads.
A conditional generative adversarial network, Proton2Carbon, was developed to translate proton pencil beam images into synthetic carbon ion beam images. The model was trained on 547 224 paired proton-carbon images acquired with a scintillation detector at the Marburg Ion Therapy Centre. Image reconstruction was performed using a 2D lateral method, and the model was evaluated on internal and external datasets for spatial resolution, using custom 3D-printed line pair modules.
The Proton2Carbon model improved the spatial resolution of pRads from 1.7 to 2.7 lp/cm on internal data and to 2.3 lp/cm on external data, demonstrating generalizability. Water equivalent thickness accuracy remained consistent with pRads and cRads. Evaluation on an anthropomorphic head phantom showed enhanced structural clarity, though some increased noise was observed.
This study demonstrates that deep learning can enhance pRad image quality by leveraging paired proton-carbon data. Proton2Carbon can be integrated into existing imaging workflows to improve clinical and research applications of proton radiography. To facilitate further research, the full dataset used to train Proton2Carbon is publicly released and available at https://zenodo.org/records/14945165.
集成模式质子成像技术是一种临床上可实现的获取质子射线照片（pRads）的方法，但其空间分辨率受多次库仑散射（MCS）限制。由于MCS的幅度随粒子电荷增加而减小，较重的离子（如碳离子）能产生分辨率更高的射线照片（cRads）。因此，通过使用经过训练的图像转换网络将单个质子笔形束图像转换为等效的碳离子数据，可能提高pRads的图像分辨率。该方法可解释为应用具有空间可变点扩散函数的数据驱动去卷积操作。
提出一种基于质子 - 碳配对数据的深度学习框架，以提高集成模式pRads的分辨率。
开发了一种条件生成对抗网络Proton2Carbon，用于将质子笔形束图像转换为合成碳离子束图像。该模型在马尔堡离子治疗中心使用闪烁探测器获取的547224对质子 - 碳配对图像上进行训练。使用二维横向方法进行图像重建，并使用定制的3D打印线对模块在内部和外部数据集上评估模型的空间分辨率。
Proton2Carbon模型将内部数据上pRads的空间分辨率从1.7提高到2.7 lp/cm，外部数据上提高到2.3 lp/cm，证明了其通用性。水等效厚度精度与pRads和cRads保持一致。在人体头部模型上的评估显示结构清晰度增强，不过观察到一些噪声增加。
本研究表明深度学习可通过利用质子 - 碳配对数据提高pRad图像质量。Proton2Carbon可集成到现有的成像工作流程中，以改善质子射线照相的临床和研究应用。为便于进一步研究，用于训练Proton2Carbon的完整数据集已公开发布，可在https://zenodo.org/records/14945165获取。
