Generalization of intensity distribution of medical images using GANs

Lee, Dong‐Ho; Li, Yan; Shin, Byeong‐Seok

doi:10.1186/s13673-020-00220-2

Cited by 12 publications

(9 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The performance of models trained on datasets with intensity distributions considerably different from those of the testing datasets is poor. 22 However, we found that the segmentation performance was comparable among the three imaging devices imaging devices regardless of different intensity distributions (Figure 4). The DSC was improved adjusting the window level and width in the preliminary experiment.…”

Section: Discussionmentioning

confidence: 84%

“…Generally, CNNs for medical images perform poorly for new medical images with intensity distributions that are completely different from those of the training dataset. 22 Furthermore, the trade-off between FOV size and resolution of the input image remains problematic for segmentation using CNN. 23 To solve these problems practically, the generalization of new datasets needs to be considered.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Evaluation of generalization ability for deep learning‐based auto‐segmentation accuracy in limited field of view CBCT of male pelvic region

Hirashima

Nakamura

Imanishi³

et al. 2023

J Applied Clin Med Phys

View full text Add to dashboard Cite

The aim of this study was to evaluate generalization ability of segmentation accuracy for limited FOV CBCT in the male pelvic region using a full-image CNN. Auto-segmentation accuracy was evaluated using various datasets with different intensity distributions and FOV sizes. Methods: A total of 171 CBCT datasets from patients with prostate cancer were enrolled. There were 151, 10, and 10 CBCT datasets acquired from Vero4DRT, TrueBeam STx, and Clinac-iX, respectively. The FOV for Vero4DRT, TrueBeam STx, and Clinac-iX was 20, 26, and 25 cm, respectively. The ROIs, including the bladder, prostate, rectum, and seminal vesicles, were manually delineated. The U 2 -Net CNN network architecture was used to train the segmentation model. A total of 131 limited FOV CBCT datasets from Vero4DRT were used for training (104 datasets) and validation (27 datasets); thereafter the rest were for testing. The training routine was set to save the best weight values when the DSC in the validation set was maximized. Segmentation accuracy was qualitatively and quantitatively evaluated between the ground truth and predicted ROIs in the different testing datasets. Results:The mean scores ± standard deviation of visual evaluation for bladder, prostate, rectum, and seminal vesicle in all treatment machines were 1.0 ± 0.7, 1.5 ± 0.6, 1.4 ± 0.6, and 2.1 ± 0.8 points, respectively. The median DSC values for all imaging devices were ≥0.94 for the bladder, 0.84-0.87 for the prostate and rectum, and 0.48-0.69 for the seminal vesicles. Although the DSC values for the bladder and seminal vesicles were significantly different among the three imaging devices, the DSC value of the bladder changed by less than 1% point. The median MSD values for all imaging devices were ≤1.2 mm for the bladder and 1.4-2.2 mm for the prostate, rectum, and seminal vesicles. The MSD values for the seminal vesicles were significantly different between the three imaging devices. Conclusion:The proposed method is effective for testing datasets with different intensity distributions and FOV from training datasets.

show abstract

Section: Discussionmentioning

confidence: 84%

Section: Introductionmentioning

confidence: 99%

Evaluation of generalization ability for deep learning‐based auto‐segmentation accuracy in limited field of view CBCT of male pelvic region

Hirashima

Nakamura

Imanishi³

et al. 2023

J Applied Clin Med Phys

View full text Add to dashboard Cite

show abstract

“…With the application of depth learning in identification [ 57 ], in addition to segmentation [ 58 ] of images and reconstruction [ 59 , 60 ], in future research, we will focus on combining the depth learning technology to further improve the accuracy of the model. In addition, we will utilize big data technologies to build the nonlinear mapping relationship between [ 61 ] stimulation parameters and stimulation effects to provide more comprehensive guidance for stimulation.…”

Section: Discussionmentioning

confidence: 99%

Chinese Digital Arm (CDA): A High-Precision Digital Arm for Electrical Stimulation Simulation

Zhang

Wang

et al. 2023

Bioengineering

View full text Add to dashboard Cite

To effectively analyze the diffusion and accumulation of signals on the surface and inside the human body under electrical stimulation, we used the gray threshold of the Chinese Digital Human image dataset to segment an arm image and reconstruct the tissue to obtain its three-dimensional cloud point dataset. Finally, a semirefined digital arm entity model with the geometric characteristics of the actual human arm tissue was constructed using reverse engineering technology. Further input of the current signal stimulation under tDCS and tACS with additional analysis of the signal diffusion in the transient mode via model calculation revealed that DC electrical stimulation is likely to cause high-voltage burns. The effective depth achieved using the AC stimulation signal is considerable, and provides reference for the electrical stimulation selection. Simultaneously, in the digital arm model, the signal diffusion and tissue damage inside the arm can be analyzed by changing the field, which provides a theoretical basis for the experimental study of the human body.

show abstract

“…With the development of generative adversarial networks (GANs) [21], Chia et al [5] utilized a GAN to propose style conversion networks that can transfer an image style from another image, which also can be applied to transferring color to a grayscale image, or to synthesize a color image from a grayscale image [10,15,28,32]. The CycleGAN [44], in particular, enables the learning of unpaired datasets by applying cycle-consistency.…”

Section: Introductionmentioning

confidence: 99%

Unpaired medical image colorization using generative adversarial network

Liang

Lee

et al. 2021

Multimed Tools Appl

Self Cite

View full text Add to dashboard Cite

We consider medical image transformation problems where a grayscale image is transformed into a color image. The colorized medical image should have the same features as the input image because extra synthesized features can increase the possibility of diagnostic errors. In this paper, to secure colorized medical images and improve the quality of synthesized images, as well as to leverage unpaired training image data, a colorization network is proposed based on the cycle generative adversarial network (CycleGAN) model, combining a perceptual loss function and a total variation (TV) loss function. Visual comparisons and experimental indicators from the NRMSE, PSNR, and SSIM metrics are used to evaluate the performance of the proposed method. The experimental results show that GAN-based style conversion can be applied to colorization of medical images. As well, the introduction of perceptual loss and TV loss can improve the quality of images produced as a result of colorization better than the result generated by only using the CycleGAN model.

show abstract

Generalization of intensity distribution of medical images using GANs

Cited by 12 publications

References 27 publications

Evaluation of generalization ability for deep learning‐based auto‐segmentation accuracy in limited field of view CBCT of male pelvic region

Evaluation of generalization ability for deep learning‐based auto‐segmentation accuracy in limited field of view CBCT of male pelvic region

Chinese Digital Arm (CDA): A High-Precision Digital Arm for Electrical Stimulation Simulation

Unpaired medical image colorization using generative adversarial network

Contact Info

Product

Resources

About