Reduction of metal artefacts in CT with Cycle-GAN

Du, Muge; Liang, Kaichao; Xing, Yuxiang

doi:10.1109/nssmic.2018.8824544

Cited by 13 publications

(12 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Two unsupervised MAR methods are investigated: an improved version of CycleGAN-based MAR [22] and ADN [25]. They are representative unsupervised MAR methods and can be trained with unpaired practical data.…”

Section: Image-domain Unsupervised Mar Methodsmentioning

confidence: 99%

“…Representative dual-domain methods include the earlier CNNMAR [12] which combines conventional NMAR [20] in sinogram domain and deep MAR in image domain, and later methods like DudoNet [13] and DudoNet++ [14] which jointly train the sinogramdomain and image-domain networks and connect them with differentiable reconstruction layer and can recover certain In this work, we investigate the domain gap problem in several deep MAR methods, which is crucial for clinical application. The investigated methods cover image-domain supervised MAR (a supervised baseline and I-DL-MAR [8]), dual-domain supervised MAR (DudoNet [13] and DudoNet++ [14]) and image-domain unsupervised MAR (CycleGAN-based MAR [22,23] and ADN [25]). The comparison is done on a dental dataset and a torso dataset which both contain simulated and practical data.…”

Section: Introductionmentioning

confidence: 99%

“…The comparison is done on a dental dataset and a torso dataset which both contain simulated and practical data. Compared to previous clinical study for evaluating unsupervised MAR methods [22,25], the two datasets are of higher resolution, and especially the dental dataset contains more severe artefacts. In addition, we analyze the causes of domain gap problem in each method based on the experimental results and give possible recommendations.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Investigation of domain gap problem in several deep-learning-based CT metal artefact reduction methods

Liang²,

Liu³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Metal artefacts in CT images may disrupt image quality and interfere with diagnosis. Recently many deep-learning-based CT metal artefact reduction (MAR) methods have been proposed. Current deep MAR methods may be troubled with domain gap problem, where methods trained on simulated data cannot perform well on practical data. In this work, we experimentally investigate two image-domain supervised methods, two dual-domain supervised methods and two image-domain unsupervised methods on a dental dataset and a torso dataset, to explore whether domain gap problem exists or is overcome. We find that I-DL-MAR and DudoNet are effective for practical data of the torso dataset, indicating the domain gap problem is solved. However, none of the investigated methods perform satisfactorily on practical data of the dental dataset. Based on the experimental results, we further analyze the causes of domain gap problem for each method and dataset, which may be beneficial for improving existing methods or designing new ones. The findings suggest that the domain gap problem in deep MAR methods remains to be addressed.

show abstract

Section: Image-domain Unsupervised Mar Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Investigation of domain gap problem in several deep-learning-based CT metal artefact reduction methods

Liang²,

Liu³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Recent studies have applied GANs to MAR in small regions of CT images of the ear [35]- [38]. Du et al [39] presented preliminary results from GAN-based MAR for images with dental fillings, while Liao et al [40] proposed a CycleGAN-based artifact disentanglement network and compared quantitative evaluation results against existing supervised/unsupervised MAR methods using synthesized datasets. However, the performance of GAN-based MAR remains to be clarified when the network is trained with clinical CT images containing complex artifacts derived from multiple dental fillings.…”

Section: Introductionmentioning

confidence: 99%

Regularized Three-Dimensional Generative Adversarial Nets for Unsupervised Metal Artifact Reduction in Head and Neck CT Images

Nakao

Imanishi²,

Imai

et al. 2020

IEEE Access

View full text Add to dashboard Cite

The reduction of metal artifacts in computed tomography (CT) images, specifically for strong artifacts generated from multiple metal objects, is a challenging issue in medical imaging research. Although there have been some studies on supervised metal artifact reduction through the learning of synthesized artifacts, it is difficult for simulated artifacts to cover the complexity of the real physical phenomena that may be observed in X-ray propagation. In this paper, we introduce metal artifact reduction methods based on an unsupervised volume-to-volume translation learned from clinical CT images. We construct three-dimensional adversarial nets with a regularized loss function designed for metal artifacts from multiple dental fillings. The results of experiments using a CT volume database of 361 patients demonstrate that the proposed framework has an outstanding capacity to reduce strong artifacts and to recover underlying missing voxels, while preserving the anatomical features of soft tissues and tooth structures from the original images.

show abstract

“…Du et al [35] presented preliminary results from GAN-based MAR for images with dental fillings, while Liao et al [36] proposed a CycleGAN-based artifact disentanglement network and compared quantitative evaluation results against existing supervised/unsupervised MAR methods using synthesized datasets. However, the performance of GAN-based MAR remains to be clarified when the network is trained with clinical CT images containing complex artifacts derived from multiple dental fillings.…”

Section: Introductionmentioning

confidence: 99%

Three-dimensional Generative Adversarial Nets for Unsupervised Metal Artifact Reduction

Nakao,

Imanishi,

Ueda

et al. 2019

Preprint

View full text Add to dashboard Cite

The reduction of metal artifacts in computed tomography (CT) images, specifically for strong artifacts generated from multiple metal objects, is a challenging issue in medical imaging research. Although there have been some studies on supervised metal artifact reduction through the learning of synthesized artifacts, it is difficult for simulated artifacts to cover the complexity of the real physical phenomena that may be observed in X-ray propagation. In this paper, we introduce metal artifact reduction methods based on an unsupervised volume-to-volume translation learned from clinical CT images. We construct threedimensional adversarial nets with a regularized loss function designed for metal artifacts from multiple dental fillings. The results of experiments using 915 CT volumes from real patients demonstrate that the proposed framework has an outstanding capacity to reduce strong artifacts and to recover underlying missing voxels, while preserving the anatomical features of soft tissues and tooth structures from the original images.

show abstract

Reduction of metal artefacts in CT with Cycle-GAN

Cited by 13 publications

References 5 publications

Investigation of domain gap problem in several deep-learning-based CT metal artefact reduction methods

Investigation of domain gap problem in several deep-learning-based CT metal artefact reduction methods

Regularized Three-Dimensional Generative Adversarial Nets for Unsupervised Metal Artifact Reduction in Head and Neck CT Images

Three-dimensional Generative Adversarial Nets for Unsupervised Metal Artifact Reduction

Contact Info

Product

Resources

About