DDPTransformer: Dual-Domain With Parallel Transformer Network for Sparse View CT Image Reconstruction

Li, Runrui; Li, Qing; Wang, Hexi; Li, Saize; Zhao, Juanjuan; Qiang, Yan; Wang, Long

doi:10.1109/tci.2022.3207351

Cited by 19 publications

(11 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We compare GloReDi with the following state-of-the-art methods: DDNet [59], FBPConvNet [22], DuDoNet [31], DDPTrans [29], and DuDoTrans [47]. In addition, we name the network without the distillation from I T as the frequency encoder and decoder network (Fred-Net), optimized with only the pixel-wise loss in Eq.…”

Section: Comparison With State-of-the-art Methodsmentioning

confidence: 99%

“…Furthermore, various techniques have been proposed to enhance dual domain approaches through network design [1,6] and unrolling architecture [58]. Recently, Transformer [46] has been introduced to dual-domain methods for its capability of capturing long-range dependencies, achieving superior performance [29,47,50]. However, the problems of the irreversible secondary artifacts and additional computational costs are not well addressed, and the requirement of raw data greatly limits their generalizability to other CT scanners/protocols.…”

Section: Deep-learning-based Sparse-view Ctmentioning

confidence: 99%

“…However, due to the limited receptive field, these methods have difficulty in modeling the global information efficiently, thus leading to suboptimal results. Current dominant methods turn to the help of sinogram by restoring the sinogram and performing CT image post-processing simultaneously [17,29,31,47]. Specifically, in addition to the image domain methods, deep neural networks are also applied to interpolate the missing data in sinogram so that the reconstructed images can be recovered in a global manner [10,25].…”

Section: Introductionmentioning

confidence: 99%

“…First, the processing of sinogram data is sensitive and even subtle changes may remarkably affect the reconstructed images and introduce stubborn secondary artifacts that are hard to be eliminated. Second, in ultra sparse-view scenarios, inpainting methods are unable to accurately restore the excessive missing raw data; e.g., inpainting sparse CT data of 18 views to the full CT data of 720 views only retains about 2.5% of the valid information, in which the involvement of sinogram processing makes the learning more difficult and compromise the performance [29,47]. Third, it is often impractical to access sinogram data given the privacy and commercial concerns, and if accessible, the requirement of raw data greatly limits the generalization to other scanners/protocols.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Transfer Learning in Motor Imagery Brain Computer Interface: A Review

2022

J. Shanghai Jiaotong Univ. (Sci.)

View full text Add to dashboard Cite

Sparse-view computed tomography (CT)-using a small number of projections for tomographic reconstructionenables much lower radiation dose to patients and accelerated data acquisition. The reconstructed images, however, suffer from strong artifacts, greatly limiting their diagnostic value. Current trends for sparse-view CT turn to the raw data for better information recovery. The resultant dual-domain methods, nonetheless, suffer from secondary artifacts, especially in ultra-sparse view scenarios, and their generalization to other scanners/protocols is greatly limited. A crucial question arises: have the image post-processing methods reached the limit? Our answer is not yet. In this paper, we stick to image post-processing methods due to great flexibility and propose global representation(GloRe) distillation framework for sparse-view CT, termed GloReDi. First, we propose to learn GloRe with Fourier convolution, so each element in GloRe has an image-wide receptive field. Second, unlike methods that only use the full-view images for supervision, we propose to distill GloRe from intermediate-view reconstructed images that are readily available but not explored in previous literature. The success of GloRe distillation is attributed to two key components: representation directional distillation to align the GloRe directions, and band-pass-specific contrastive distillation to gain clinically important details. Extensive experiments demonstrate the superiority of the proposed GloReDi over the state-of-the-art methods, including dual-domain ones. The source code is available at https: //github.com/longzilicart/GloReDi.

show abstract

Section: Comparison With State-of-the-art Methodsmentioning

confidence: 99%

Section: Deep-learning-based Sparse-view Ctmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Transfer Learning in Motor Imagery Brain Computer Interface: A Review

2022

J. Shanghai Jiaotong Univ. (Sci.)

View full text Add to dashboard Cite

show abstract

“…However, these methods only consider single-domain information and ignore the consistency of projection data. To remedy this issue, a large number of dual-domain SVCT networks with cascaded or parallel single-domain sub-network modules have been proposed [51][52][53][54][55][56]. These methods have achieved promising results in the field of sparse-view reconstruction, but none of them consider the presence of metallic implants.…”

Section: Svctmentioning

confidence: 99%

Mud-Net: multi-domain deep unrolling network for simultaneous sparse-view and metal artifact reduction in computed tomography

Shi,

Jiang,

Zhang

et al. 2024

Mach. Learn.: Sci. Technol.

View full text Add to dashboard Cite

Sparse-view computed tomography (SVCT) is regarded as a promising technique to accelerate data acquisition and reduce radiation dose. However, in the presence of metallic implants, SVCT inevitably makes the reconstructed CT images suffer from severe metal artifacts and streaking artifacts due to the lack of sufficient projection data. Previous stand-alone SVCT and metal artifact reduction (MAR) methods to solve the problem of simultaneously sparse-view and metal artifact reduction (SVMAR) are plagued by insufficient correction accuracy. To overcome this limitation, we propose a multi-domain deep unrolling network, called Mud-Net, for SVMAR. Specifically, we establish a joint sinogram, image, artifact, and coding domains deep unrolling reconstruction model to recover high-quality CT images from the under-sampled sinograms corrupted by metallic implants. To train this multi-domain network effectively, we embed multi-domain knowledge into the network training process. Comprehensive experiments demonstrate that our method is superior to both existing MAR methods in the full-view MAR task and previous SVCT methods in the SVMAR task.

show abstract

Self-supervised learning for CT image denoising and reconstruction: a review

Choi

2024

Biomed. Eng. Lett.

View full text Add to dashboard Cite

DDPTransformer: Dual-Domain With Parallel Transformer Network for Sparse View CT Image Reconstruction

Cited by 19 publications

References 38 publications

Transfer Learning in Motor Imagery Brain Computer Interface: A Review

Transfer Learning in Motor Imagery Brain Computer Interface: A Review

Mud-Net: multi-domain deep unrolling network for simultaneous sparse-view and metal artifact reduction in computed tomography

Self-supervised learning for CT image denoising and reconstruction: a review

Contact Info

Product

Resources

About