Gajewski Jan, Borys Damian, De Gregorio Angelica, Kopeć Renata, Krah Nils, Krzempek Dawid, Patera Vincenzo, Rinaldi Ilaria, Rydygier Marzena, Sarrut David, Schiavi Angelo, Skóra Tomasz, Stasica-Dudek Paulina, Rucinski Antoni
Institute of Nuclear Physics Polish Academy of Sciences, Krakow, PL-31342, Poland.
Silesian University of Technology, Department of Systems Biology and Engineering, Gliwice, Poland.
Comput Biol Med. 2025 Sep;196(Pt B):110802. doi: 10.1016/j.compbiomed.2025.110802. Epub 2025 Jul 29.
This work aims at implementation, validation, and proof of application of fast, voxel-based single proton linear energy transfer (LET) spectra scoring for clinical proton therapy. The LET spectra provide more comprehensive information on the mixed radiation field produced by protons in heterogeneous patient geometry in comparison to the dose-averaged LET (LET), commonly investigated pre-clinically and clinically.
We implemented single particle spectra scoring methods for LET and other physics quantities, e.g. deposited energy or track length, characterising mixed radiation fields in voxelized geometries. The scorers were implemented in a GPU-accelerated Monte Carlo (MC) code Fred, as well as a general purpose MC codes Gate/Geant4 and Fluka. The validation included a comparison of spectra obtained with Fred, Gate and Fluka, and evaluating the calculation performance. The LET spectra were also calculated for an intensity-modulated proton therapy (IMPT) patient treatment plan and compared to LET.
Implementation of spectra scorers of various quantities, including the LET was shown to be conducted accurately and spectra obtained with Fred, Gate and Fluka are in excellent agreement. The GPU acceleration allows precise and time-efficient calculation of the LET spectra and can be conducted within about an hour with Fred (using two GPU cards) compared to tens of hours with Gate or Fluka (using up to 400 CPUs). We have also shown that for an IMPT patient treatment plan, single particle LET spectra may differ for the same LET, being potentially responsible for uncertainties in the advanced treatment planning methods based on variable RBE and LET.
Single particle LET spectra scoring is possible with the state-of-the-art Monte Carlo methods, allowing a more detailed insight into radiation effects in mixed radiation fields produced by proton beams in a human body and, thanks to time efficient calculations enables future clinical translation of the method.
本研究旨在实现、验证并证明基于体素的快速单质子线性能量传递（LET）谱评分在临床质子治疗中的应用。与临床前和临床常用研究的剂量平均LET相比，LET谱能提供更多关于质子在非均匀患者几何结构中产生的混合辐射场的综合信息。
我们实现了用于LET及其他物理量（如沉积能量或径迹长度）的单粒子谱评分方法，以表征体素化几何结构中的混合辐射场。这些评分器在GPU加速的蒙特卡罗（MC）代码Fred以及通用MC代码Gate/Geant4和Fluka中实现。验证包括比较用Fred、Gate和Fluka获得的谱，并评估计算性能。还针对调强质子治疗（IMPT）患者治疗计划计算了LET谱，并与LET进行比较。
包括LET在内的各种物理量的谱评分器的实现被证明是准确的，并且用Fred、Gate和Fluka获得的谱高度一致。GPU加速使得LET谱的计算精确且高效，使用Fred（使用两张GPU卡）大约一小时即可完成，而使用Gate或Fluka（使用多达400个CPU）则需要数十小时。我们还表明，对于IMPT患者治疗计划，相同LET下的单粒子LET谱可能不同，这可能是基于可变相对生物效应（RBE）和LET的先进治疗计划方法存在不确定性的原因。
使用先进的蒙特卡罗方法可以实现单粒子LET谱评分，从而更深入地了解质子束在人体中产生的混合辐射场中的辐射效应，并且由于计算效率高，该方法有望在未来实现临床应用。
