SkrGAN: Sketching-Rendering Unconditional Generative Adversarial Networks for Medical Image Synthesis

Zhang, Tianyang; Fu, Huazhu; Zhao, Yitian; Cheng, Jun; Guo, Mengjie; Gu, Zaiwang; Yang, Bing; Xiao, Yuting; Gao, Shenghua; Liu, Jiang

doi:10.1007/978-3-030-32251-9_85

Cited by 51 publications

(25 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…With the adversarial learning between the G and D, the G is promoted to generate artificial images bearing greater similarity to real images. PGGANs is an extension to the GAN training process, and achieves high-resolution synthetization (in this study, 256 × 256 pixels) by alternating training and adding new networks G and D. To generate OCT images with correct anatomic structures, a sketch guidance modules G that contains the edge and detailed information was added to growing networks G. 13 With each additional layer, the resolution is increased (e.g., from 4 × 4 to 8 × 8) allowing the generation of higher-resolution images. All of the GAN development and experiments were conducted using PyTorch (version 1.0, Facebook, USA).…”

Section: Methodsmentioning

confidence: 99%

Assessment of Generative Adversarial Networks Model for Synthetic Optical Coherence Tomography Images of Retinal Disorders

Zheng

Xie

Zhou

et al. 2020

Trans. Vis. Sci. Tech.

Self Cite

View full text Add to dashboard Cite

Purpose To assess whether a generative adversarial network (GAN) could synthesize realistic optical coherence tomography (OCT) images that satisfactorily serve as the educational images for retinal specialists, and the training datasets for the classification of various retinal disorders using deep learning (DL). Methods The GANs architecture was adopted to synthesize high-resolution OCT images trained on a publicly available OCT dataset, including urgent referrals (37,206 OCT images from eyes with choroidal neovascularization, and 11,349 OCT images from eyes with diabetic macular edema) and nonurgent referrals (8617 OCT images from eyes with drusen, and 51,140 OCT images from normal eyes). Four hundred real and synthetic OCT images were evaluated by two retinal specialists (with over 10 years of clinical retinal experience) to assess image quality. We further trained two DL models on either real or synthetic datasets and compared the performance of urgent versus nonurgent referrals diagnosis tested on a local (1000 images from the public dataset) and clinical validation dataset (278 images from Shanghai Shibei Hospital). Results The image quality of real versus synthetic OCT images was similar as assessed by two retinal specialists. The accuracy of discrimination of real versus synthetic OCT images was 59.50% for retinal specialist 1 and 53.67% for retinal specialist 2. For the local dataset, the DL model trained on real (DL_Model_R) and synthetic OCT images (DL_Model_S) had an area under the curve (AUC) of 0.99, and 0.98, respectively. For the clinical dataset, the AUC was 0.94 for DL_Model_R and 0.90 for DL_Model_S. Conclusions The GAN synthetic OCT images can be used by clinicians for educational purposes and for developing DL algorithms. Translational Relevance The medical image synthesis based on GANs is promising in humans and machines to fulfill clinical tasks.

show abstract

Section: Methodsmentioning

confidence: 99%

Assessment of Generative Adversarial Networks Model for Synthetic Optical Coherence Tomography Images of Retinal Disorders

Zheng

Xie

Zhou

et al. 2020

Trans. Vis. Sci. Tech.

Self Cite

View full text Add to dashboard Cite

show abstract

“…Zhang et al 155 . proposed a sketching‐rendering unconditional GAN (SkrGAN), which can not only augment grayscale images, such as X‐Ray, CT and MRI, but also colour images such as retinal colour fundus photography for various segmentation tasks.…”

Section: Methodsmentioning

confidence: 99%

A review of medical image data augmentation techniques for deep learning applications

et al. 2021

View full text Add to dashboard Cite

Summary Research in artificial intelligence for radiology and radiotherapy has recently become increasingly reliant on the use of deep learning‐based algorithms. While the performance of the models which these algorithms produce can significantly outperform more traditional machine learning methods, they do rely on larger datasets being available for training. To address this issue, data augmentation has become a popular method for increasing the size of a training dataset, particularly in fields where large datasets aren’t typically available, which is often the case when working with medical images. Data augmentation aims to generate additional data which is used to train the model and has been shown to improve performance when validated on a separate unseen dataset. This approach has become commonplace so to help understand the types of data augmentation techniques used in state‐of‐the‐art deep learning models, we conducted a systematic review of the literature where data augmentation was utilised on medical images (limited to CT and MRI) to train a deep learning model. Articles were categorised into basic, deformable, deep learning or other data augmentation techniques. As artificial intelligence models trained using augmented data make their way into the clinic, this review aims to give an insight to these techniques and confidence in the validity of the models produced.

show abstract

“…By providing the target data as an additional input, a conditional GAN directs the data generation process as it is constrained by the target data. This additional constraint makes conditional GAN more suitable for image translation than an unconditional GAN (Zhang et al ., 2019). While the input data for a conditional GAN can be either classes or images, in this paper, the focus is on conditional image GANs, specifically the Pix2Pix GAN.…”

Section: Methodsmentioning

confidence: 99%

Pre‐migration diffraction separation using generative adversarial networks

Lowney

Lokmer

O’Brien

et al. 2021

Geophysical Prospecting

View full text Add to dashboard Cite

Diffraction imaging is the process of separating diffraction events from the seismic wavefield and imaging them independently, highlighting subsurface discontinuities. While there are many analytic‐based methods for diffraction imaging which use kinematic, dynamic or both, properties of the diffracted wavefield, they can be slow and require parameterization. Here, we propose an image‐to‐image generative adversarial network to automatically separate diffraction events on pre‐migrated seismic data in a fraction of the time of conventional methods. To train the generative adversarial network, plane‐wave destruction was applied to a range of synthetic and real images from field data to create training data. These training data were screened and any areas where the plane‐wave destruction did not perform well, such as synclines and areas of complex dip, were removed to prevent bias in the neural network. A total of 14,132 screened images were used to train the final generative adversarial network. The trained network has been applied across several geologically distinct field datasets, including a 3D example. Here, generative adversarial network separation is shown to be comparable to a benchmark separation created with plane‐wave destruction, and up to 12 times faster. This demonstrates the clear potential in generative adversarial networks for fast and accurate diffraction separation.

show abstract

SkrGAN: Sketching-Rendering Unconditional Generative Adversarial Networks for Medical Image Synthesis

Cited by 51 publications

References 14 publications

Assessment of Generative Adversarial Networks Model for Synthetic Optical Coherence Tomography Images of Retinal Disorders

Assessment of Generative Adversarial Networks Model for Synthetic Optical Coherence Tomography Images of Retinal Disorders

A review of medical image data augmentation techniques for deep learning applications

Pre‐migration diffraction separation using generative adversarial networks

Contact Info

Product

Resources

About