Department of Radiation Oncology, The University of California, San Francisco, San Francisco, California, USA.
Department of Radiology and Biomedical Imaging, The University of California, San Francisco, San Francisco, California, USA.
Med Phys. 2022 Oct;49(10):6622-6634. doi: 10.1002/mp.15876. Epub 2022 Aug 8.
Megavoltage computed tomography (MVCT) has been implemented on many radiotherapy treatment machines for on-board anatomical visualization, localization, and adaptive dose calculation. Implementing an MR-only workflow by synthesizing MVCT from magnetic resonance imaging (MRI) would offer numerous advantages for treatment planning and online adaptation.
In this work, we sought to synthesize MVCT (sMVCT) datasets from MRI using deep learning to demonstrate the feasibility of MRI-MVCT only treatment planning.
MVCTs and T1-weighted MRIs for 120 patients treated for head-and-neck cancer were retrospectively acquired and co-registered. A deep neural network based on a fully-convolutional 3D U-Net architecture was implemented to map MRI intensity to MVCT HU. Input to the model were volumetric patches generated from paired MRI and MVCT datasets. The U-Net was initialized with random parameters and trained on a mean absolute error (MAE) objective function. Model accuracy was evaluated on 18 withheld test exams. sMVCTs were compared to respective MVCTs. Intensity-modulated volumetric radiotherapy (IMRT) plans were generated on MVCTs of four different disease sites and compared to plans calculated onto corresponding sMVCTs using the gamma metric and dose-volume-histograms (DVHs).
MAE values between sMVCT and MVCT datasets were 93.3 ± 27.5, 78.2 ± 27.5, and 138.0 ± 43.4 HU for whole body, soft tissue, and bone volumes, respectively. Overall, there was good agreement between sMVCT and MVCT, with bone and air posing the greatest challenges. The retrospective dataset introduced additional deviations due to sinus filling or tumor growth/shrinkage between scans, differences in external contours due to variability in patient positioning, or when immobilization devices were absent from diagnostic MRIs. Dose distributions of IMRT plans evaluated for four test cases showed close agreement between sMVCT and MVCT images when evaluated using DVHs and gamma dose metrics, which averaged to 98.9 ± 1.0% and 96.8 ± 2.6% analyzed at 3%/3 mm and 2%/2 mm, respectively.
MVCT datasets can be generated from T1-weighted MRI using a 3D deep convolutional neural network with dose calculation on a sample sMVCT in close agreement with the MVCT. These results demonstrate the feasibility of using MRI-derived sMVCT in an MR-only treatment planning workflow.
为了实现在线解剖可视化、定位和自适应剂量计算，许多放射治疗设备都配备了兆伏级 CT（MVCT）。通过从磁共振成像（MRI）合成 MVCT，可以为治疗计划和在线自适应提供许多优势。
本研究旨在使用深度学习从 MRI 合成 MVCT（sMVCT）数据集，以验证仅基于 MRI-MVCT 进行治疗计划的可行性。
回顾性获取了 120 例头颈部癌症患者的 MVCT 和 T1 加权 MRI，并进行了配准。基于全卷积 3D U-Net 架构的深度神经网络用于将 MRI 强度映射到 MVCT HU。该模型的输入是从配对的 MRI 和 MVCT 数据集生成的体积贴片。U-Net 采用随机参数初始化，并在均方误差（MAE）目标函数上进行训练。使用 18 个保留的测试集评估模型的准确性。将 sMVCT 与各自的 MVCT 进行比较。在四个不同疾病部位的 MVCT 上生成调强适形容积放疗（IMRT）计划，并使用伽马度量和剂量体积直方图（DVH）将其与相应的 sMVCT 上计算的计划进行比较。
sMVCT 与 MVCT 数据集之间的 MAE 值分别为全身、软组织和骨容积的 93.3±27.5、78.2±27.5 和 138.0±43.4 HU。总体而言，sMVCT 与 MVCT 之间具有良好的一致性，骨骼和空气是最大的挑战。由于扫描之间鼻窦填充或肿瘤生长/收缩、由于患者定位的可变性导致的外部轮廓差异，或者当诊断性 MRI 中没有固定装置时，回顾性数据集引入了额外的偏差。对于四个测试案例评估的 IMRT 计划的剂量分布，当使用 DVH 和伽马剂量度量进行评估时，sMVCT 和 MVCT 图像之间显示出非常好的一致性，在 3%/3mm 和 2%/2mm 时分别平均为 98.9±1.0%和 96.8±2.6%。
使用 3D 深度卷积神经网络可以从 T1 加权 MRI 生成 MVCT 数据集，在使用样本 sMVCT 进行剂量计算时，与 MVCT 非常吻合。这些结果表明，在仅基于 MRI 的治疗计划工作流程中使用 MRI 衍生的 sMVCT 是可行的。
