Synthesis of pseudo-CT images from pelvic MRI images based on an MD-CycleGAN model for radiotherapy

Abstract

Objective: A multi-discriminator-based cycle generative adversarial network (MD-CycleGAN) model was proposed to synthesize higher-quality pseudo-CT from MRI. Approach: The MRI and CT images obtained at the simulation stage with cervical cancer were selected to train the model. The generator adopted the DenseNet as the main architecture. The local and global discriminators based on convolutional neural network jointly discriminated the authenticity of the input image data. In the testing phase, the model was verified by four-fold cross-validation method. In the prediction stage, the data were selected to evaluate the accuracy of the pseudo-CT in anatomy and dosimetry, and they were compared with the pseudo-CT synthesized by GAN with generator based on the architectures of ResNet, sU-Net, and FCN. Main results: There are significant differences(P<0.05) in the four-fold-cross validation results on peak signal-to-noise ratio and structural similarity index metrics between the pseudo-CT obtained based on MD-CycleGAN and the ground truth CT (CTgt). The pseudo-CT synthesized by MD-CycleGAN had closer anatomical information to the CTgt with root mean square error of 47.83±2.92 HU and normalized mutual information value of 0.9014±0.0212 and mean absolute error value of 46.79±2.76 HU. The differences in dose distribution between the pseudo-CT obtained by MD-CycleGAN and the CTgt were minimal. The mean absolute dose errors of Dosemax, Dosemin and Dosemean based on the planning target volume were used to evaluate the dose uncertainty of the four pseudo-CT. The u-values of the Wilcoxon test were 55.407, 41.82 and 56.208, and the differences were statistically significant. The 2%/2 mm-based gamma pass rate (%) of the proposed method was 95.45±1.91, and the comparison methods (ResNet_GAN, sUnet_GAN and FCN_GAN) were 93.33±1.20, 89.64±1.63 and 87.31±1.94, respectively. Significance: The pseudo-CT obtained based on MD-CycleGAN have higher imaging quality and are closer to the CTgt in terms of anatomy and dosimetry than other GAN models.

Improving synthetic CT accuracy by combining the benefits of multiple normalized preprocesses

“…The CycleGAN model is an unsupervised network that has been widely used to generate synthetic CT images. 5,7,10,[26][27][28] The CycleGAN architecture is consistent with that of the original paper described by Zhu et al (2017), which includes two generators (Gmvcbctct and Gct-mvcbct) and two discriminators (Dct and Dmvcbct), using ResUNet by Xiao et al (2018) and 70×70 PatchGAN described by Zhu et al (2017) as generators and discriminators, respectively. 29,30 Gmvcbct-ct converts MVCBCT to CT images and Gct-mvcbct converts CT to MVCBCT images.…”

“…The CycleGAN model is an unsupervised network that has been widely used to generate synthetic CT images 5,7,10,26–28 . The CycleGAN architecture is consistent with that of the original paper described by Zhu et al.…”

GAN‐based metal artifacts region inpainting in brain MRI imaging with reflective registration

“…Additionally, with the increasing popularity of MRI accelerators, researchers are also exploring pseudo-CT synthesis from MRI for radiotherapy planning design. 7 More satisfactory CT images can be generated by using inpainted MRI. However, current inpainting methods can only inpaint missing data using surrounding information, and the inpainting effect is not always satisfactory.…”

“…Inpainted MRI may aid in better matching the template, thus estimating a more accurate attenuation image. Additionally, with the increasing popularity of MRI accelerators, researchers are also exploring pseudo‐CT synthesis from MRI for radiotherapy planning design 7 . More satisfactory CT images can be generated by using inpainted MRI.…”

“…MRI and CT are often required for registration in radiotherapy, and metal artifacts in MRI will also affect the registration results 5 . The generation of pseudo‐CT from MRI for adaptive radiotherapy has also become a hot spot in clinical research 6,7 . The metal artifacts in MRI will seriously affect the image quality and accuracy of the generated CT. Consequently, reducing and avoiding the effects of metal artifacts is a high‐priority problem.…”