3D radiotherapy dose prediction on head and neck cancer patients with a hierarchically densely connected U-net deep learning architecture
3D radiotherapy dose prediction on head and neck cancer patients with a hierarchically densely connected U-net deep learning architecture
复制标题
DOI：
10.1088/1361-6560/ab039b
复制
发表时间：
2019-03-01
影响因子：
3.5
通讯作者：
Jiang, Steve
中科院分区：
文献类型：
--
作者：
Dan Nguyen;Jia, Xun;Jiang, Steve
由于靶区体积大、处方剂量水平多以及靶区附近有许多对辐射敏感的关键结构，头颈 (H&N) 癌症患者的治疗计划过程被认为是最复杂的过程之一。该部位的治疗计划需要高水平的人类专业知识和巨大的努力来制定个性化的高质量计划，需要长达一周的时间，这会降低肿瘤控制和患者生存的机会。为了解决这个问题，我们建议研究一种基于深度学习的剂量预测模型，即分层密集连接 U-net，它基于两种非常流行的网络架构：U-net 和 DenseNet。我们发现这种新架构能够准确有效地预测剂量分布，在测试数据的均匀性、剂量一致性和剂量覆盖方面优于其他两种模型（标准 U-net 和 DenseNet）。对所有处于风险的器官进行平均，我们提出的模型能够根据测试数据预测处于风险的器官最大剂量在处方剂量的 6.3% 以内，平均剂量在处方剂量的 5.1% 以内。其他模型，Standard U-net 和 DenseNet，表现较差，平均危险器官最大剂量预测误差分别为 8.2% 和 9.3%，平均平均剂量预测误差分别为 6.4% 和 6.8%。此外，我们提出的模型使用的可训练参数比标准 U 网少 12 倍，并且预测患者剂量的速度比 DenseNet 快 4 倍。
The treatment planning process for patients with head and neck (H&N) cancer is regarded as one of the most complicated due to large target volume, multiple prescription dose levels, and many radiation-sensitive critical structures near the target. Treatment planning for this site requires a high level of human expertise and a tremendous amount of effort to produce personalized high quality plans, taking as long as a week, which deteriorates the chances of tumor control and patient survival. To solve this problem, we propose to investigate a deep learning-based dose prediction model, Hierarchically Densely Connected U-net, based on two highly popular network architectures: U-net and DenseNet. We find that this new architecture is able to accurately and efficiently predict the dose distribution, outperforming the other two models, the Standard U-net and DenseNet, in homogeneity, dose conformity, and dose coverage on the test data. Averaging across all organs at risk, our proposed model is capable of predicting the organ-at-risk max dose within 6.3% and mean dose within 5.1% of the prescription dose on the test data. The other models, the Standard U-net and DenseNet, performed worse, having an averaged organ-at-risk max dose prediction error of 8.2% and 9.3%, respectively, and averaged mean dose prediction error of 6.4% and 6.8%, respectively. In addition, our proposed model used 12 times less trainable parameters than the Standard U-net, and predicted the patient dose 4 times faster than DenseNet.