National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, 17 Panjiayuannanli, Chaoyang District, Beijing, 100021, China.
Med Phys. 2020 Mar;47(3):1249-1257. doi: 10.1002/mp.14004. Epub 2020 Jan 28.
The purpose of this study is to develop a deep learning (DL) method for producing four-dimensional computed tomography (4DCT) ventilation imaging and to evaluate the accuracy of the DL-based ventilation imaging against single-photon emission-computed tomography (SPECT) ventilation imaging (SPECT-VI). The performance of the DL-based method is assessed by comparing with density change- and Jacobian-based (HU and JAC) methods.
Fifty patients with esophagus or lung cancer who underwent thoracic radiotherapy were enrolled in this study. For each patient, 4DCT scans paired with 99mTc-Technegas SPECT/CT were acquired before the first radiotherapy treatment. 4DCT and SPECT/CT were first rigidly registered using MIMvista and converted to data matrix using MATLAB, and then transferred to a DL model based on U-net for correlating 4DCT features and SPECT-VI. Two forms of 4DCT dataset [(a) ten phases and (b) two phases of peak-exhalation and peak-inhalation] as input are studied. Tenfold cross-validation procedure was used to evaluate the performance of the DL model. For comparative evaluation, HU and JAC methodologies are used to calculate specific ventilation imaging based on 4DCT (CTVI) for each patient. The voxel-wise Spearman's correlation was evaluated over the whole lung between each of CTVI and corresponding SPECT-VI. The SPECT-VI and produced CTVIs were segmented into high, median, and low functional lung (HFL, MFL, and LFL) regions. The spatial overlap of corresponding HFL, MFL, and LFL for each CTVI against SPECT-VI was also evaluated using the dice similarity coefficient (DSC). The averaged DSC of functional lung regions was calculated and statistically analyzed with a one-factor ANONA model among different methods.
The voxel-wise Spearman r values were (0.22 ± 0.31), (-0.09 ± 0.18), and (0.73 ± 0.16)/(0.71 ± 0.17) for the CTVI , CTVI , and CTVI /CTVI . These results showed the DL method yielded the strongest correlation with SPECT-VI. Using the DSC as the spatial overlap metric, we found that the CTVI , CTVI , and CTVI /CTVI methods achieved averaged DSC values for all patients to be (0.45 ± 0.08), (0.33 ± 0.04), and (0.73 ± 0.09)/(0.71 ± 0.09), respectively. The results demonstrated that the DL method yielded the highest similarity with SPECT-VI with the prominently significant difference (P < 10 ).
This study developed a DL method for producing CTVI and performed a validation against SPECT-VI. The results demonstrated that DL method can derive CTVI with greatly improved accuracy in comparison to HU and JAC methods. The produced ventilation images can be more accurate and useful for lung functional avoidance radiotherapy and treatment response modeling.
本研究旨在开发一种用于生成四维计算机断层扫描（4DCT）通气成像的深度学习（DL）方法，并评估基于 DL 的通气成像与单光子发射计算机断层扫描（SPECT）通气成像（SPECT-VI）的准确性。通过与密度变化和雅可比（HU 和 JAC）方法的比较，评估基于 DL 的方法的性能。
本研究纳入了 50 名患有食管癌或肺癌的患者，他们接受了胸部放疗。对于每位患者，在首次放疗前采集了 4DCT 扫描和 99mTc-Technegas SPECT/CT 配对。首先使用 MIMvista 对 4DCT 和 SPECT/CT 进行刚性配准，并使用 MATLAB 将其转换为数据矩阵，然后将其传输到基于 U-net 的 DL 模型中，以关联 4DCT 特征和 SPECT-VI。研究了两种形式的 4DCT 数据集[（a）十个相位和（b）呼气峰和吸气峰的两个相位]作为输入。使用 10 倍交叉验证程序评估 DL 模型的性能。为了进行比较评估，HU 和 JAC 方法用于基于每个患者的 4DCT（CTVI）计算特定的通气成像。在整个肺部对每个 CTVI 和相应的 SPECT-VI 进行了体素水平的 Spearman 相关分析。将 SPECT-VI 和生成的 CTVIs 分割为高、中、低功能肺（HFL、MFL 和 LFL）区域。还使用骰子相似系数（DSC）评估了每个 CTVI 与 SPECT-VI 之间对应 HFL、MFL 和 LFL 的空间重叠。计算了功能肺区的平均 DSC，并使用单因素方差分析模型对不同方法进行了统计分析。
CTVI、CTVI 和 CTVI / CTVI 的体素水平 Spearman r 值分别为（0.22 ± 0.31）、（-0.09 ± 0.18）和（0.73 ± 0.16）/（0.71 ± 0.17）。这些结果表明，DL 方法与 SPECT-VI 具有最强的相关性。使用 DSC 作为空间重叠度量，我们发现 CTVI、CTVI 和 CTVI / CTVI 方法对所有患者的平均 DSC 值分别为（0.45 ± 0.08）、（0.33 ± 0.04）和（0.73 ± 0.09）/（0.71 ± 0.09）。结果表明，DL 方法与 SPECT-VI 具有最高的相似性，具有显著差异（P < 10）。
本研究开发了一种用于生成 CTVI 的 DL 方法，并与 SPECT-VI 进行了验证。结果表明，与 HU 和 JAC 方法相比，DL 方法可以极大地提高 CTVI 的准确性。生成的通气图像可以更准确和有用，用于肺功能避免放疗和治疗反应建模。
