Department of Radiology, Konkuk University Medical Center, Seoul, Korea.
Department of Radiology, Seoul National University Bundang Hospital, Seongnam, Korea.
Korean J Radiol. 2020 Mar;21(3):356-364. doi: 10.3348/kjr.2019.0413.
To compare the image quality of low-dose (LD) computed tomography (CT) obtained using a deep learning-based denoising algorithm (DLA) with LD CT images reconstructed with a filtered back projection (FBP) and advanced modeled iterative reconstruction (ADMIRE).
One hundred routine-dose (RD) abdominal CT studies reconstructed using FBP were used to train the DLA. Simulated CT images were made at dose levels of 13%, 25%, and 50% of the RD (DLA-1, -2, and -3) and reconstructed using FBP. We trained DLAs using the simulated CT images as input data and the RD CT images as ground truth. To test the DLA, the American College of Radiology CT phantom was used together with 18 patients who underwent abdominal LD CT. LD CT images of the phantom and patients were processed using FBP, ADMIRE, and DLAs (LD-FBP, LD-ADMIRE, and LD-DLA images, respectively). To compare the image quality, we measured the noise power spectrum and modulation transfer function (MTF) of phantom images. For patient data, we measured the mean image noise and performed qualitative image analysis. We evaluated the presence of additional artifacts in the LD-DLA images.
LD-DLAs achieved lower noise levels than LD-FBP and LD-ADMIRE for both phantom and patient data (all < 0.001). LD-DLAs trained with a lower radiation dose showed less image noise. However, the MTFs of the LD-DLAs were lower than those of LD-ADMIRE and LD-FBP (all < 0.001) and decreased with decreasing training image dose. In the qualitative image analysis, the overall image quality of LD-DLAs was best for DLA-3 (50% simulated radiation dose) and not significantly different from LD-ADMIRE. There were no additional artifacts in LD-DLA images.
DLAs achieved less noise than FBP and ADMIRE in LD CT images, but did not maintain spatial resolution. The DLA trained with 50% simulated radiation dose showed the best overall image quality.
比较基于深度学习的去噪算法（DLA）获得的低剂量（LD）计算机断层扫描（CT）与滤波反投影（FBP）和高级模型迭代重建（ADMIRE）重建的 LD CT 图像的图像质量。
使用 FBP 重建了 100 例常规剂量（RD）腹部 CT 研究，以训练 DLA。模拟 CT 图像以 RD 的 13%、25%和 50%剂量水平制作（DLA-1、-2 和-3），并使用 FBP 重建。我们使用模拟 CT 图像作为输入数据和 RD CT 图像作为真实数据来训练 DLAs。为了测试 DLA，使用美国放射学院 CT 体模和 18 名接受腹部 LD CT 的患者。使用 FBP、ADMIRE 和 DLAs（分别为 LD-FBP、LD-ADMIRE 和 LD-DLA 图像）处理体模和患者的 LD CT 图像。为了比较图像质量，我们测量了体模图像的噪声功率谱和调制传递函数（MTF）。对于患者数据，我们测量了平均图像噪声并进行了定性图像分析。我们评估了 LD-DLA 图像中是否存在额外的伪影。
对于体模和患者数据，LD-DLAs 的噪声水平均低于 LD-FBP 和 LD-ADMIRE（均<0.001）。用较低辐射剂量训练的 LD-DLAs 显示出较低的图像噪声。然而，LD-DLAs 的 MTF 低于 LD-ADMIRE 和 LD-FBP（均<0.001），并随训练图像剂量的降低而降低。在定性图像分析中，LD-DLAs 的整体图像质量对于 DLA-3（50%模拟辐射剂量）最佳，与 LD-ADMIRE 无显著差异。LD-DLA 图像中没有额外的伪影。
与 FBP 和 ADMIRE 相比，DLAs 在 LD CT 图像中实现了更低的噪声，但没有保持空间分辨率。用 50%模拟辐射剂量训练的 DLA 显示出最佳的整体图像质量。
