Department of Radiation Oncology, Stanford University, Stanford, 94305-5847, CA, USA; Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, Guangdong 518055, China.
College of Information and Communication Engineering, Communication University of China, Beijing 100024, China.
Comput Biol Med. 2021 Nov;138:104775. doi: 10.1016/j.compbiomed.2021.104775. Epub 2021 Aug 21.
Software-based methods can improve CT spatial resolution without changing the hardware of the scanner or increasing the radiation dose to the object. In this work, we aim to develop a deep learning (DL) based CT super-resolution (SR) method that can reconstruct low-resolution (LR) sinograms into high-resolution (HR) CT images. We mathematically analyzed imaging processes in the CT SR imaging problem and synergistically integrated the SR model in the sinogram domain and the deblur model in the image domain into a hybrid model (SADIR). SADIR incorporates the CT domain knowledge and is unrolled into a DL network (SADIR-Net). The SADIR-Net is a self-supervised network, which can be trained and tested with a single sinogram. SADIR-Net was evaluated through SR CT imaging of a Catphan physical phantom and a real porcine phantom, and its performance was compared to the other state-of-the-art (SotA) DL-based CT SR methods. On both phantoms, SADIR-Net obtains the highest information fidelity criterion (IFC), structure similarity index (SSIM), and lowest root-mean-square-error (RMSE). As to the modulation transfer function (MTF), SADIR-Net also obtains the best result and improves the MTF by 69.2% and MTF by 69.5% compared with FBP. Alternatively, the spatial resolutions at MTF and MTF from SADIR-Net can reach 91.3% and 89.3% of the counterparts reconstructed from the HR sinogram with FBP. The results show that SADIR-Net can provide performance comparable to the other SotA methods for CT SR reconstruction, especially in the case of extremely limited training data or even no data at all. Thus, the SADIR method could find use in improving CT resolution without changing the hardware of the scanner or increasing the radiation dose to the object.
基于软件的方法可以在不改变扫描仪硬件或增加物体辐射剂量的情况下提高 CT 空间分辨率。在这项工作中，我们旨在开发一种基于深度学习（DL）的 CT 超分辨率（SR）方法，该方法可以将低分辨率（LR）正弦图重建为高分辨率（HR）CT 图像。我们从数学上分析了 CT SR 成像问题中的成像过程，并协同地将 SR 模型集成到正弦图域和去模糊模型集成到图像域中，形成一个混合模型（SADIR）。SADIR 结合了 CT 领域的知识，并被展开成一个 DL 网络（SADIR-Net）。SADIR-Net 是一个自监督网络，可以用单个正弦图进行训练和测试。SADIR-Net 通过对 Catphan 物理体模和真实猪体模的 SR CT 成像进行了评估，并将其性能与其他最先进的（SoTA）基于 DL 的 CT SR 方法进行了比较。在这两个体模上，SADIR-Net 获得了最高的信息保真度准则（IFC）、结构相似性指数（SSIM）和最低均方根误差（RMSE）。至于调制传递函数（MTF），SADIR-Net 也获得了最好的结果，与 FBP 相比，MTF 提高了 69.2%，MTF 提高了 69.5%。或者，SADIR-Net 的 MTF 和 MTF 空间分辨率可以达到 FBP 重建的 HR 正弦图对应值的 91.3%和 89.3%。结果表明，SADIR-Net 可以为 CT SR 重建提供与其他 SoTA 方法相当的性能，特别是在训练数据极其有限甚至没有数据的情况下。因此，SADIR 方法可以在不改变扫描仪硬件或增加物体辐射剂量的情况下提高 CT 分辨率。
