发表机构
Umeå University; Medical University of Vienna（于默奥大学; 维也纳医科大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本研究提出PyDoseRT Photon，一种结合神经先验与残差校正的物理引导笔形束剂量引擎，用于CT和MRI的光子剂量计算，在DoseRAD2026任务中取得低剂量误差与高伽马通过率，且运行时间满足要求。
AI 中文摘要
我们提出一种混合的、基于物理的分析型笔形束（PB）剂量引擎，该引擎由两个小型冻结神经物理先验增强，随后接一个3D卷积残差校正网络（U-Net）。我们使用PyDoseRT Photon解决DoseRAD2026光子剂量预测任务，PyDoseRT Photon是一个在PyTorch中实现的GPU加速PB引擎，在三个物理特异性递减的学习阶段中向蒙特卡洛（MC）精度校正。该引擎在可行区域精确复现挑战赛的无头MC源，并使用以射束通量加权放射深度评估的射束质量索引笔形核，以及在相互作用位点的TERMA源缩放来建模患者。两个小型神经先验通过冻结引擎训练后保持冻结：一个3.9万参数的2D通量校正，以及一个48参数的横向异质性校正，该校正混合了质量守恒的高斯重分布算子。一个紧凑的3D U-Net（136万参数）带有顺序细化分支，随后从七个通道中为射束视线（BEV）帧中的每个控制点（CP）预测有界乘性增益和加性残差。所有学习阶段均从零初始化，因此训练从解析解开始。对于MRI，一个nnU-Net回归模型合成CT，该CT进入相同的流程，依次使用与CT分支相同的训练校正器。设计选择由挑战赛排名驱动，其中运行时间权重翻倍。所提交方法在我们的本地CT和MR验证数据集上评估，CP MAE分别为0.0086和0.0098，IDD距离分别为0.0011和0.0013，计划MAE分别为0.0024和0.0047，伽马通过率（1%/1mm）分别为99.12%和96.95%，运行时间分别为46秒和49秒。
英文摘要
We present a hybrid, physics-based analytical pencil-beam (PB) dose engine, augmented by two small frozen neural physics priors, followed by a 3-D convolutional residual-correction network (U-Net). We address the DoseRAD2026 (https://doserad2026.grand-challenge.org/) photon dose-prediction task with PyDoseRT Photon, a GPU PB engine implemented in PyTorch and corrected toward Monte Carlo (MC) accuracy in three learned stages of decreasing physical specificity. The engine reproduces the challenge's head-less MC source exactly where it can and models the patient with a beam-quality-indexed pencil kernel evaluated at the field's fluence-weighted radiological depth, and TERMA source scaling at the interaction site. Two tiny neural priors are trained through the frozen engine and then frozen themselves: a 39k-parameter 2-D fluence correction and a 48-parameter lateral heterogeneity correction mixing mass-conserving Gaussian redistribution operators. A compact 3-D U-Net (1.36M parameters) with a sequential refinement branch then predicts, per control point (CP) in the beam's-eye-view (BEV) frame, a bounded multiplicative gain and additive residual from seven channels. All learned stages are zero-initialized, so training starts from the analytical solution. For MRI, an nnU-Net regression model synthesizes a CT that enters the identical pipeline, with consecutively, the same trained corrector as the CT track. Design choices were driven by the challenge ranking, in which runtime carries double weight. The submitted method evaluated on our local CT and MR validation dataset achieved CP MAE 0.0086 and 0.0098, IDD distance 0.0011 and 0.0013, plan MAE 0.0024 and 0.0047, gamma pass rate (1%/1mm) 99.12% and 96.95%, with runtimes of 46s and 49s, respectively.
