Chu Hung, de Vette Suzanne P M, Neh Hendrike, Sijtsema Nanna M, Steenbakkers Roel J H M, Moreno Amy, Langendijk Johannes A, van Ooijen Peter M A, Fuller Clifton D, van Dijk Lisanne V
Department of Radiation Oncology, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands.
Department of Radiation Oncology, The University of Texas MD Anderson Cancer Center, Houston, Texas.
Int J Radiat Oncol Biol Phys. 2025 Jan 1;121(1):269-280. doi: 10.1016/j.ijrobp.2024.07.2334. Epub 2024 Aug 13.
Conventional normal tissue complication probability (NTCP) models for patients with head and neck cancer are typically based on single-value variables, which, for radiation-induced xerostomia, are baseline xerostomia and mean salivary gland doses. This study aimed to improve the prediction of late xerostomia by using 3-dimensional information from radiation dose distributions, computed tomography imaging, organ-at-risk segmentations, and clinical variables with deep learning (DL).
An international cohort of 1208 patients with head and neck cancer from 2 institutes was used to train and twice validate DL models (deep convolutional neural network, EfficientNet-v2, and ResNet) with 3-dimensional dose distribution, computed tomography scan, organ-at-risk segmentations, baseline xerostomia score, sex, and age as input. The NTCP endpoint was moderate-to-severe xerostomia 12 months postradiation therapy. The DL models' prediction performance was compared with a reference model: a recently published xerostomia NTCP model that used baseline xerostomia score and mean salivary gland doses as input. Attention maps were created to visualize the focus regions of the DL predictions. Transfer learning was conducted to improve the DL model performance on the external validation set.
All DL-based NTCP models showed better performance (area under the receiver operating characteristic curve [AUC], 0.78-0.79) than the reference NTCP model (AUC, 0.74) in the independent test. Attention maps showed that the DL model focused on the major salivary glands, particularly the stem cell-rich region of the parotid glands. DL models obtained lower external validation performance (AUC, 0.63) than the reference model (AUC, 0.66). After transfer learning on a small external subset, the DL model (AUC, 0.66) performed better than the reference model (AUC, 0.64).
DL-based NTCP models performed better than the reference model when validated in data from the same institute. Improved performance in the external data set was achieved with transfer learning, demonstrating the need for multicenter training data to realize generalizable DL-based NTCP models.
头颈部癌患者的传统正常组织并发症概率（NTCP）模型通常基于单值变量，对于放射性口干症而言，这些变量为基线口干症和平均唾液腺剂量。本研究旨在通过使用来自放射剂量分布、计算机断层扫描成像、危及器官分割和临床变量的三维信息以及深度学习（DL）来改进晚期口干症的预测。
来自2个机构的1208名头颈部癌患者的国际队列用于训练和两次验证DL模型（深度卷积神经网络、EfficientNet-v2和ResNet），输入包括三维剂量分布、计算机断层扫描、危及器官分割、基线口干症评分、性别和年龄。NTCP终点为放疗后12个月的中重度口干症。将DL模型的预测性能与参考模型进行比较：一个最近发表的口干症NTCP模型，该模型使用基线口干症评分和平均唾液腺剂量作为输入。创建注意力图以可视化DL预测的关注区域。进行迁移学习以提高DL模型在外部验证集上的性能。
在独立测试中，所有基于DL的NTCP模型均表现出比参考NTCP模型（AUC为0.74）更好的性能（受试者操作特征曲线下面积[AUC]为0.78 - 0.79）。注意力图显示DL模型聚焦于主要唾液腺，特别是腮腺中富含干细胞的区域。DL模型在外部验证中的性能（AUC为0.63）低于参考模型（AUC为0.66）。在一个小的外部子集上进行迁移学习后，DL模型（AUC为0.66）的表现优于参考模型（AUC为0.64）。
在来自同一机构的数据中进行验证时，基于DL的NTCP模型表现优于参考模型。通过迁移学习在外部数据集中实现了性能提升，这表明需要多中心训练数据来实现可推广的基于DL的NTCP模型。
