CaGAN: A Cycle-Consistent Generative Adversarial Network With Attention for Low-Dose CT Imaging

Huang, Zhiyuan; Chen, Zixiang; Zhang, Qiyang; Quan, Guotao; Ji, Min; Zhang, Chengjin; Yang, Yongfeng; Liu, Xin; Liang, Dong; Zheng, Hairong; Hu, Zhanli

doi:10.1109/tci.2020.3012928

Cited by 61 publications

(41 citation statements)

References 61 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Since CT images are single-channel grayscale images, the training database may not contain many valid images. One of the most common experimental strategies in LDCT noise removal tasks is to generate overlapping patches (32), which better represent the local characteristics of the image and increase the number of training samples (56). Considering the limited number of CT images that could meet the requirements of our experiment, we additionally adopted the overlapping patch strategy, which not only considers spatial interconnection between patches but also significantly accelerates the convergence of the learning model (52) Net algorithm are designed for natural images, the training process is more inclined to learn the characteristic information of natural images, and the experimental effect on grayscale images is not good.…”

Section: Discussionmentioning

confidence: 99%

“…Expanding convolution in the image domain has a sizeable sensory field that better captures correlations between anatomical regions and synthesizes lost anatomical information in the event of high signal distortion. Inspired by the successful computer vision application, Huang et al (32) introduced the attention mechanism into the cycle-consistent generative adversarial network's (CycleGAN) generator for LDCT denoise and achieved satisfactory results. In addition, Ataei et al (33) cascaded 2 identical neural networks to recreate fine structural details in low-contrast areas by minimizing perceived loss.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Low-dose computed tomography image reconstruction via a multistage convolutional neural network with autoencoder perceptual loss network

Li¹,

Li²,

Li³

et al. 2022

Quant Imaging Med Surg

View full text Add to dashboard Cite

Background: Computed tomography (CT) is widely used in medical diagnoses due to its ability to noninvasively detect the internal structures of the human body. However, CT scans with normal radiation doses can cause irreversible damage to patients. The radiation exposure is reduced with low-dose CT (LDCT), although considerable speckle noise and streak artifacts in CT images and even structural deformation may result, significantly undermining its diagnostic capability.Methods: This paper proposes a multistage network framework which gradually divides the entire process into 2-staged sub-networks to complete the task of image reconstruction. Specifically, a dilated residual convolutional neural network (DRCNN) was used to denoise the LDCT image. Then, the learned context information was combined with the channel attention subnet, which retains local information, to preserve the structural details and features of the image and textural information. To obtain recognizable characteristic details, we introduced a novel self-calibration module (SCM) between the 2 stages to reweight the local features, which realizes the complementation of information at different stages while refining feature information. In addition, we also designed an autoencoder neural network, using a self-supervised learning scheme to train a perceptual loss neural network specifically for CT images.Results: We evaluated the diagnostic quality of the results and performed ablation experiments on the loss function and network structure modules to verify each module's effectiveness in the network. Our proposed network architecture obtained high peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM), and visual information fidelity (VIF) values in terms of quantitative evaluation. In the analysis of qualitative results, our network structure maintained a better balance between eliminating image noise and preserving image details. Experimental results showed that our proposed network structure obtained better metrics and visual evaluation.Conclusions: This study proposed a new LDCT image reconstruction method by combining autoencoder perceptual loss networks with multistage convolutional neural networks (MSCNN). Experimental results showed that the newly proposed method has performance than other methods.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Low-dose computed tomography image reconstruction via a multistage convolutional neural network with autoencoder perceptual loss network

Li¹,

Li²,

Li³

et al. 2022

Quant Imaging Med Surg

View full text Add to dashboard Cite

show abstract

“…Meanwhile, the attention mechanism in GAN facilitates GAN to generate images that fulfill the clinical medical standards. Previous studies have shown [29][30][31][32][33] that channel-wise attention in GAN can be used to reconstruct more details in MR images than other methods without channel-wise attention. However, the spatial attention has been neglected by these approaches.…”

Section: E Spatial and Channel-wise Attentionmentioning

confidence: 99%

A Modified Generative Adversarial Network Using Spatial and Channel-Wise Attention for CS-MRI Reconstruction

Wang

2021

IEEE Access

View full text Add to dashboard Cite

Compressed sensing (CS) can speed up the magnetic resonance imaging (MRI) process and reconstruct high-quality images from under-sampled k-space data. However, traditional CS-MRI suffers from slow iterations and artifacts caused by noise when the acceleration factor is high. Currently, deep learning has been introduced to address these issues. Although some improvements have been achieved, the reconstruction problem under high under-sampling rates has not been solved. Thus, our study proposed a novel CS-MRI reconstruction method called RSCA-GAN. The generator of RSCA-GAN is a residual U-net consisting of both spatial and channel-wise attention. The proposed RSCA-GAN was compared to the zero-filling, DAGAN, RefineGAN, and RCA-GAN using both cartesian and non-cartesian undersampling masks on brain and knee datasets. The sampling rates of cartesian masks are 25%, 16.7%, and 12.5% and the sampling rates of spiral and radial masks are 30%, 20%, and 10%. Peak signal-to-noise ratio (PSNR), structural similarity (SSIM) and normalized mean square error (NMSE) were used to evaluate the reconstructed image quality, and the rank-sum test was adopted to evaluate the significant difference among different approaches. P < 0.05 indicated statistical significance. The results demonstrated that RSCA-GAN outperforms the other approaches for all the quantitative metrics. Thus, RSCA-GAN exhibits excellent reconstruction performance at high under-sampling rates and is suitable for clinical applications.

show abstract

“…거기에 더해 배치 정규화 (batch normalization)[11] 이 주어진 모델의 성능을 더욱 신장시킬 수 있음을 확인했다. Ca-GAN[7] 에서는 생성적 적대 신경망 (generative adversarial network (GAN)) 을…”

unclassified

“…또한 Table1에서와같이 합성 데이터 실험인 case 1과 2의 경우 4번의 교차검증 역시 같은 표본들로 검증했 다. 비교 방법은 지도 학습 기반인 DnCNN[6], RED-CNN[27],CaGAN[7]을 선택했다. 비지도 학습으로서는 본 연구의 기초가 되는 Quan et al[5] ISCL[1]과 최신 짝지어지지 않는 영상 기반 노이즈 제거 방법인 UIDNet[28], ADN[17]을 선택했다.…”

unclassified

Optimized Normalization for Unsupervised Learning-based Image Denoising

Lee¹,

Jeong²

2021

J Korea Comput Graph Soc

View full text Add to dashboard Cite

Recently, deep learning-based denoising approaches have been actively studied. In particular, with the advances of blind denoising techniques, it become possible to train a deep learning-based denoising model only with noisy images in an image domain where it is impossible to obtain a clean image. We no longer require pairs of a clean image and a noisy image to obtain a restored clean image from the observation. However, it is difficult to recover the target using a deep learning-based denoising model trained by only noisy images if the distribution of the noisy image is far from the distribution of the clean image. To address this limitation, unpaired image denoising approaches have recently been studied that can learn the denoising model from unpaired data of the noisy image and the clean image. ISCL showed comparable performance close to that of supervised learning-based models based on pairs of clean and noisy images. In this study, we propose suitable normalization techniques for each purpose of architectures (e.g., generator, discriminator, and extractor) of ISCL. We demonstrate that the proposed method outperforms state-of-the-art unpaired image denoising approaches including ISCL.

show abstract

CaGAN: A Cycle-Consistent Generative Adversarial Network With Attention for Low-Dose CT Imaging

Cited by 61 publications

References 61 publications

Low-dose computed tomography image reconstruction via a multistage convolutional neural network with autoencoder perceptual loss network

Low-dose computed tomography image reconstruction via a multistage convolutional neural network with autoencoder perceptual loss network

A Modified Generative Adversarial Network Using Spatial and Channel-Wise Attention for CS-MRI Reconstruction

Optimized Normalization for Unsupervised Learning-based Image Denoising

Contact Info

Product

Resources

About