Nishio Mizuho, Muramatsu Chisako, Noguchi Shunjiro, Nakai Hirotsugu, Fujimoto Koji, Sakamoto Ryo, Fujita Hiroshi
Department of Diagnostic Imaging and Nuclear Medicine, Kyoto University Graduate School of Medicine, 54 Shogoin Kawaharacho, Sakyo-ku, Kyoto, 606-8507, Japan.
Faculty of Data Science, Shiga University, 1-1-1 Banba, Hikone, Shiga, 522-8522, Japan.
Comput Biol Med. 2020 Nov;126:104032. doi: 10.1016/j.compbiomed.2020.104032. Epub 2020 Oct 7.
To develop and evaluate a three-dimensional (3D) generative model of computed tomography (CT) images of lung nodules using a generative adversarial network (GAN). To guide the GAN, lung nodule size was used.
A public CT dataset of lung nodules was used, from where 1182 lung nodules were obtained. Our proposed GAN model used masked 3D CT images and nodule size information to generate images. To evaluate the generated CT images, two radiologists visually evaluated whether the CT images with lung nodule were true or generated, and the diagnostic ability was evaluated using receiver-operating characteristic analysis and area under the curves (AUC). Then, two models for classifying nodule size into five categories were trained, one using the true and the other using the generated CT images of lung nodules. Using true CT images, the classification accuracy of the sizes of the true lung nodules was calculated for the two classification models.
The sensitivity, specificity, and AUC of the two radiologists were respectively as follows: radiologist 1: 81.3%, 37.7%, and 0.592; radiologist 2: 77.1%, 30.2%, and 0.597. For categorization of nodule size, the mean accuracy of the classification model constructed with true CT images was 85% (range 83.2-86.1%), and that with generated CT images was 85% (range 82.2-88.1%).
Our results show that it was possible to generate 3D CT images of lung nodules that could be used to construct a classification model of lung nodule size without true CT images.
使用生成对抗网络（GAN）开发并评估肺结节计算机断层扫描（CT）图像的三维（3D）生成模型。为指导GAN，使用了肺结节大小。
使用了一个公开的肺结节CT数据集，从中获取了1182个肺结节。我们提出的GAN模型使用带掩码的3D CT图像和结节大小信息来生成图像。为评估生成的CT图像，两名放射科医生直观评估带有肺结节的CT图像是真实的还是生成的，并使用接收者操作特征分析和曲线下面积（AUC）评估诊断能力。然后，训练了两个将结节大小分为五类的模型，一个使用真实的肺结节CT图像，另一个使用生成的肺结节CT图像。使用真实的CT图像，计算两个分类模型对真实肺结节大小的分类准确率。
两名放射科医生的敏感度、特异度和AUC分别如下：放射科医生1：81.3%、37.7%和0.592；放射科医生2：77.1%、3%和0.597。对于结节大小分类，用真实CT图像构建的分类模型的平均准确率为85%（范围83.2 - 86.1%），用生成的CT图像构建的分类模型的平均准确率为85%（范围82.2 - 88.1%）。
我们的结果表明，有可能生成可用于构建肺结节大小分类模型的肺结节3D CT图像，而无需真实的CT图像。 （注：原文中“放射科医生2：77.1%、3%和0.597”这里特异度“3%”疑似有误，根据前文推测可能是“30.2%”，翻译时保留了原文数据。）
