ICTEAM, UCLouvain, Louvain-la-Neuve, 1348, Belgium.
IREC/MIRO, UCLouvain, Brussels, 1200, Belgium.
Med Phys. 2019 Dec;46(12):5790-5798. doi: 10.1002/mp.13856. Epub 2019 Nov 1.
Monte Carlo (MC) algorithms offer accurate modeling of dose calculation by simulating the transport and interactions of many particles through the patient geometry. However, given their random nature, the resulting dose distributions have statistical uncertainty (noise), which prevents making reliable clinical decisions. This issue is partly addressable using a huge number of simulated particles but is computationally expensive as it results in significantly greater computation times. Therefore, there is a trade-off between the computation time and the noise level in MC dose maps. In this work, we address the mitigation of noise inherent to MC dose distributions using dilated U-Net - an encoder-decoder-styled fully convolutional neural network, which allows fast and fully automated denoising of whole-volume dose maps.
We use mean squared error (MSE) as loss function to train the model, where training is done in 2D and 2.5D settings by considering a number of adjacent slices. Our model is trained on proton therapy MC dose distributions of different tumor sites (brain, head and neck, liver, lungs, and prostate) acquired from 35 patients. We provide the network with input MC dose distributions simulated using particles while keeping particles as reference.
After training, our model successfully denoises new MC dose maps. On average (averaged over five patients with different tumor sites), our model recovers of 55.99 Gy from the noisy MC input of 49.51 Gy, whereas the low noise MC (reference) offers 56.03 Gy. We observed a significant reduction in average RMSE (thresholded >10% max ref) for reference vs denoised (1.25 Gy) than reference vs input (16.96 Gy) leading to an improvement in signal-to-noise ratio (ISNR) by 18.06 dB. Moreover, the inference time of our model for a dose distribution is less than 10 s vs 100 min (MC simulation using particles).
We propose an end-to-end fully convolutional network that can denoise Monte Carlo dose distributions. The networks provide comparable qualitative and quantitative results as the MC dose distribution simulated with particles, offering a significant reduction in computation time.
蒙特卡罗（MC）算法通过模拟许多粒子在患者几何结构中的传输和相互作用，提供了对剂量计算的精确建模。然而，由于其随机性，得到的剂量分布具有统计不确定性（噪声），这使得无法做出可靠的临床决策。这个问题可以通过使用大量模拟粒子部分解决，但由于计算量很大，计算时间也会显著增加。因此，MC 剂量图的计算时间和噪声水平之间存在权衡。在这项工作中，我们使用扩张 U-Net 来解决 MC 剂量分布固有的噪声问题 - 一种编码器-解码器样式的全卷积神经网络，它可以快速且全自动地对整个体积剂量图进行去噪。
我们使用均方误差（MSE）作为损失函数来训练模型，其中通过考虑多个相邻切片，在 2D 和 2.5D 设置中进行训练。我们的模型是在来自 35 名患者的不同肿瘤部位（脑、头颈部、肝、肺和前列腺）的质子治疗 MC 剂量分布上进行训练的。我们为网络提供了使用 粒子模拟的输入 MC 剂量分布，而将 粒子保留为参考。
经过训练，我们的模型成功地对新的 MC 剂量图进行了去噪。平均而言（在五个具有不同肿瘤部位的患者中平均），我们的模型从噪声较大的 MC 输入中恢复了 55.99 Gy，而低噪声 MC（参考）提供了 56.03 Gy。我们观察到平均 RMSE（阈值 > 10% max ref）从参考到去噪（1.25 Gy）的显著降低，而从参考到输入（16.96 Gy）的降低不明显，这导致信噪比（ISNR）提高了 18.06 dB。此外，与使用 粒子进行 MC 模拟相比，我们的模型对剂量分布的推断时间不到 10 秒（100 分钟）。
我们提出了一种端到端的全卷积网络，可以对 MC 剂量分布进行去噪。该网络提供了与使用 粒子模拟的 MC 剂量分布相当的定性和定量结果，同时大大减少了计算时间。
