Removing Ring Artifacts in Cbct Images Via Generative Adversarial Network

Shuyang, Zhao; Li, Jianwu; Huo, Qirun

doi:10.1109/icassp.2018.8462316

Cited by 17 publications

(11 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The resulting sCT images preserve CT image quality while keeping CBCT anatomy (18)(19)(20)(21)(22)(23)(24)(25)(26). These CBCT-to-CT translation-based methods can be generally divided into two categories: the paired (18)(19)(20) and unpaired (21)(22)(23)(24)(25)(26). Paired CBCT-to-CT translation-based methods usually involve specific loss terms for improving network performance, such as smooth loss (18) and unidirectional relative total variation loss (19).…”

Section: Introductionmentioning

confidence: 99%

“…These CBCT-to-CT translation-based methods can be generally divided into two categories: the paired (18)(19)(20) and unpaired (21)(22)(23)(24)(25)(26). Paired CBCT-to-CT translation-based methods usually involve specific loss terms for improving network performance, such as smooth loss (18) and unidirectional relative total variation loss (19). Nevertheless, these methods require paired data for model training.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

CBCT-based synthetic CT generation using generative adversarial networks with disentangled representation

Liu¹,

Yan²,

Cheng³

et al. 2021

Quant Imaging Med Surg

View full text Add to dashboard Cite

Background: Cone-beam computed tomography (CBCT) plays a key role in image-guided radiotherapy (IGRT), however its poor image quality limited its clinical application. In this study, we developed a deeplearning based approach to translate CBCT image to synthetic CT (sCT) image that preserves both CT image quality and CBCT anatomical structures.Methods: A novel synthetic CT generative adversarial network (sCTGAN) was proposed for CBCT-to-CT translation via disentangled representation. The approach of disentangled representation was employed to extract the anatomical information shared by CBCT and CT image domains. Both on-board CBCT and planning CT of 40 patients were used for network learning and those of another 12 patients were used for testing. Accuracy of our network was quantitatively evaluated using a series of statistical metrics, including the peak signal-to-noise ratio (PSNR), mean structural similarity index (SSIM), mean absolute error (MAE), and root-mean-square error (RMSE). Effectiveness of our network was compared against three state-of-theart CycleGAN-based methods.Results: The PSNR, SSIM, MAE, and RMSE between sCT generated by sCTGAN and deformed planning CT (dpCT) were 34.12 dB, 0.86, 32.70 HU, and 60.53 HU, while the corresponding values between original CBCT and dpCT were 28.67 dB, 0.64, 70.56 HU, and 112.13 HU. The RMSE (60.53±14.38 HU) of sCT generated by sCTGAN was less than that of sCT generated by all the three comparing methods (72.40±16.03 HU by CycleGAN, 71.60±15.09 HU by CycleGAN-Unet512, 64.93±14.33 HU by CycleGAN-AG). Conclusions:The sCT generated by our sCTGAN network was closer to the ground truth (dpCT), in comparison to all the three comparing CycleGAN-based methods. It provides an effective way to generate high-quality sCT which has a wide application in IGRT and adaptive radiotherapy.

show abstract

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

CBCT-based synthetic CT generation using generative adversarial networks with disentangled representation

Liu¹,

Yan²,

Cheng³

et al. 2021

Quant Imaging Med Surg

View full text Add to dashboard Cite

show abstract

“…Among them, the first column is the original images, the second column is the corrected images, the third column is the residual images, and the fourth column is the zoomed image details. We compare the proposed algorithm with several state-of-the-art ring removal methods via sinogram domain, including polar coordinate based method [9], wavelet-Fourier correction [13], GAN-based method [11], median filtering-based method [19], and polynomial interpolation method [20], and the correction images are shown in Fig. 6, of which all images are set with the same window width and level, and the first column in each row is in original size, the 2 nd and 3 rd column are the enlarged part of the original one.…”

Section: Resultsmentioning

confidence: 99%

“…Xiao et al proposed a method [9] for extracting ring artifacts in polar coordinates, though the image resolution is kept well, but it may not be suitable for the complicated scenario as it supposes that the ring artifacts in the θ direction have the same gray value. The dictionary representation [10] and the deep learning based methods [11] are also used in reconstructed images. These methods achieve better results than before, though the problem of image blurring still exist.…”

Section: Introductionmentioning

confidence: 99%

Ring Artifacts Reduction in CBCT: Pixels Detection and Patch Based Correction

Zhao¹,

Li²,

Luo³

2019

Molecular &Amp; Cellular Biomechanics

View full text Add to dashboard Cite

The ring artifacts introduced by the defective pixels with non-linear responses in the high-resolution detector, have a great impact on subsequent processing and quantitative analysis of the reconstructed images. In this paper, a multistep method is proposed to suppress the ring artifacts of micro CT images, which firstly locates the positions of the defective pixels in the sinogram, and then corrects the corresponding value in the projections. Since the defective pixels always appear as vertical stripes in the sinogram, a horizontal curve is derived by summing the pixel values along vertical direction, thus the abrupt segments related to the defective stripes are enhanced notably, and a proportion coefficient based on the second derivative of the curve is taken as the indicator for the position and the severity of the defective pixels. Then, the detected defective pixels in the sinogram are transferred and relocated in the projections, an improved 3D block matching filtering (BM3D) algorithm is applied to restore the defective pixels in corresponding projection images. In the end, the tomographic images are reconstructed from the corrected projections. In the experiment, a small piece of the motherwort's rhizome and a part of a mouse's lung are imaged by micro-CT, and the result shows that, compared with the other four state-of-art methods, the proposed method has a great reduction on the ring artifacts of the reconstructed images, and makes less impact in spatial resolution and contrast in the same time.

show abstract

“…In the study of (Khatter et al), the authors have applied a multiscale retinex mechanism over CBCT to perform a precise assessment of root canal anatomy for endodontic therapy [61]. An image pre-processing I2I scheme based on neural network architecture is adopted in the research work of (Zhao et al), which considers generative adversarial networks (GAN) to suppress ring artifacts [62]. Mean-shift algorithm-based image segmentation is adopted in the study of (Gunawan et al).…”

Section: State-of-the-art Reviewsmentioning

confidence: 99%

A Survey on Dental Imaging for Building Classifier to Benefit the Dental Implant Practitioners

Shashikala¹,

Thangadurai²

2021

IJACSA

View full text Add to dashboard Cite

Endo-osseous implants are considered an ideal dental fixture. It is becoming the preferred choice of the edentulous patient to rehabilitate toothlessness because of their aesthetic and functional outcome. Despite the successful surgery and implant placement, complications occur, which may be related to several factors, like operative assessment, treatment planning, patient-related factors, surgical procedures, and surgeons' experience. Comprehensive radiological assessment plays a vital role in clinical analysis for better treatment planning, avoiding complications, and increasing the Implant's success rate. However, despite the variety of dental imaging, choosing the right imaging technology has become difficult for clinical experts. The investigative survey conducted in this paper aims to determine the correlation between different imaging modalities, their essential role in implant therapy. This review extensively discussed which types of computational operations applied to image modalities in the existing literature address various noises and other relevant issues. These study findings reveal significant issues with various dental imaging modalities and provide an understanding to bridge all existing research gaps towards building cost-effective classification and predictive models for accurate dental treatment planning and higher implant success rates.

show abstract

Removing Ring Artifacts in Cbct Images Via Generative Adversarial Network

Cited by 17 publications

References 12 publications

CBCT-based synthetic CT generation using generative adversarial networks with disentangled representation

CBCT-based synthetic CT generation using generative adversarial networks with disentangled representation

Ring Artifacts Reduction in CBCT: Pixels Detection and Patch Based Correction

A Survey on Dental Imaging for Building Classifier to Benefit the Dental Implant Practitioners

Contact Info

Product

Resources

About