Department of Health Technology and Informatics, The Hong Kong Polytechnic University, Hong Kong.
Department of Health Technology and Informatics, The Hong Kong Polytechnic University, Hong Kong; Department of Nuclear Medicine, Queen Mary Hospital, Hong Kong.
Int J Radiat Oncol Biol Phys. 2021 Aug 1;110(5):1508-1518. doi: 10.1016/j.ijrobp.2021.02.032. Epub 2021 Mar 6.
Our purpose was to develop a deep learning-based computed tomography (CT) perfusion mapping (DL-CTPM) method that synthesizes lung perfusion images from CT images.
This paper presents a retrospective analysis of the pulmonary technetium-99m-labeled macroaggregated albumin single-photon emission CT (SPECT)/CT scans obtained from 73 patients at Queen Mary Hospital in Hong Kong in 2019. The left and right lung scans were separated to double the size of the data set to 146. A 3-dimensional attention residual neural network was constructed to extract textural features from the CT images and reconstruct corresponding functional images. Eighty-four samples were randomly selected for training and cross-validation, and the remaining 62 were used for model testing in terms of voxel-wise agreement and function-wise concordance. To assess the voxel-wise agreement, the Spearman's correlation coefficient (R) and structural similarity index measure between the images predicted by the DL-CTPM and the corresponding SPECT perfusion images were computed to assess the statistical and perceptual image similarities, respectively. To assess the function-wise concordance, the Dice similarity coefficient (DSC) was computed to determine the similarity of the low/high functional lung volumes.
The evaluation of the voxel-wise agreement showed a moderate-to-high voxel value correlation (0.6733 ± 0.1728) and high structural similarity (0.7635 ± 0.0697) between the SPECT and DL-CTPM predicted perfusions. The evaluation of the function-wise concordance obtained an average DSC value of 0.8183 ± 0.0752 for high-functional lungs (range, 0.5819-0.9255) and 0.6501 ± 0.1061 for low-functional lungs (range, 0.2405-0.8212). Ninety-four percent of the test cases demonstrated high concordance (DSC >0.7) between the high-functional volumes contoured from the predicted and ground-truth perfusions.
We developed a novel DL-CTPM method for estimating perfusion-based lung functional images from the CT domain using a 3-dimensional attention residual neural network, which yielded moderate-to-high voxel-wise approximations of lung perfusion. To further contextualize these results toward future clinical application, a multi-institutional large-cohort study is warranted.
我们旨在开发一种基于深度学习的计算机断层扫描（CT）灌注成像（DL-CTPM）方法，该方法可从 CT 图像合成肺灌注图像。
本研究对 2019 年在香港玛丽医院进行的 73 例肺锝-99m 标记大聚合白蛋白单光子发射 CT（SPECT）/CT 扫描进行回顾性分析。将左右肺扫描分开，将数据集大小加倍至 146 个。构建一个 3 维注意力残差神经网络，从 CT 图像中提取纹理特征，并重建相应的功能图像。随机选择 84 个样本进行训练和交叉验证，其余 62 个样本用于模型测试，以评估体素级别的一致性和功能级别的一致性。为了评估体素级别的一致性，计算了由 DL-CTPM 预测的图像与相应的 SPECT 灌注图像之间的 Spearman 相关系数（R）和结构相似性指数度量，以分别评估统计和感知图像相似性。为了评估功能级别的一致性，计算了 Dice 相似系数（DSC）以确定低/高功能肺容积的相似性。
体素级别的评估显示，SPECT 和 DL-CTPM 预测的灌注之间具有中等至高的体素值相关性（0.6733 ± 0.1728）和高结构相似性（0.7635 ± 0.0697）。功能级别的评估获得了高功能肺的平均 DSC 值为 0.8183 ± 0.0752（范围为 0.5819-0.9255）和低功能肺的平均 DSC 值为 0.6501 ± 0.1061（范围为 0.2405-0.8212）。94%的测试案例显示，从预测和真实灌注中勾画的高功能容积之间具有高一致性（DSC>0.7）。
我们开发了一种新颖的基于深度学习的 CT 肺灌注功能图像估算方法，该方法使用 3 维注意力残差神经网络从 CT 域估算肺灌注功能图像，该方法可对肺灌注进行中等至高的体素级近似。为了将这些结果进一步应用于未来的临床应用，需要进行多机构大样本研究。
