Learning domain transfer for unsupervised magnetic resonance imaging restoration and edge enhancement

Fu, Bo; Dong, Yuhan; Fu, Shilin; Mao, Yuanxin; Thanh, Dang N. H.

doi:10.1002/ima.22658

Cited by 10 publications

(3 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Conventional model-based approaches have been commonly used for blind deblurring. Most of these deblurring studies have utilised regularisation techniques to estimate blur kernel, and considerable attention has been paid to estimate the regularisation term that best describes prior of natural images [15,76,92,93]. However, such optimisation techniques are not efficient due to various noises that can drastically affect estimated kernel quality, an iterative framework that requires prior information based on some statistical assumptions, and many more.…”

Section: Discussionmentioning

confidence: 99%

Multi‐scale GAN with residual image learning for removing heterogeneous blur

Khan

Luo

2022

IET Image Processing

View full text Add to dashboard Cite

Processing images with heterogeneous blur remains challenging due to multiple degradation aspects that could affect structural properties. This study proposes a deep learning‐based multi‐scaled generative adversarial network (GAN) with residual image learning to process variant and in‐variant blur. Different scaled images with corresponding gradients are concatenated as a multi‐channel single input for the proposed GAN. Residual‐ and dense‐networks are combined to explore salient features in the bottleneck section while addressing the vanishing gradient problem. A hybrid content loss function with a gradient penalty minimises the error between generated and ground truth images. Due to structure sparsity, the generated output may lose some information that leads to artifacts. Residual image learning with dilation and end‐to‐end training is used to resolve this issue by recovering high‐resolution anatomical details. Three different datasets: GoPro, Köhler, and Lai, with variant and in‐variant blur, are used to perform qualitative and quantitative analyses. Experiments show the proposed method is effective in reducing blur while preserving structural properties compared to multiple preprocessing techniques for image analysis. Moreover, the consistently improved performance over multiple publicly available datasets validates the merits of the proposed method for large data analysis.

show abstract

Section: Discussionmentioning

confidence: 99%

Multi‐scale GAN with residual image learning for removing heterogeneous blur

Khan

Luo

2022

IET Image Processing

View full text Add to dashboard Cite

show abstract

“…With the introduced pixel-level NSS prior, they propose an accurate noise level estimation method, and then develop a blind image denoising method based on the lifting Haar transform and Wiener filtering techniques. For the restoration of real medical images with multiple degradation factors, an unsupervised learning method based on domain transfer to resolve the MRI restoration problem is proposed [96]. Specifically, learning invariant representations from a degraded image and a transferred image via adversarial domain adaption and then extra selfsupervised modules to further improve the robustness of representations are used.…”

Section: Priori Denoisingmentioning

confidence: 99%

A Novel Gray Image Denoising Method Using Convolutional Neural Network

Meng

Zhang

2022

IEEE Access

View full text Add to dashboard Cite

In order to make the image denoising more effective in high noise level environment, a gray image denoising method based on symmetric dilation residual network is proposed for additive Gaussian white noise. First, the incremental & dilated convolution, BN (Batch Normalization) and leaky ReLU (Rectified Linear Unit) are used to extract and learn features from the input noise image. Then the image is reconstructed by combining the descending & dilated convolution and ReLU. Finally, the effective separation between image and noise is realized by integrating deep residual network and BN. Experimental data of specific denoising model and stochastic denoising model show that this method achieves better objective results based on small resolution input & large resolution input and small dataset & large dataset compared with other methods. A series of subjective visual comparison results also show that proposed method can perform image denoising well in high noise level environment, and there are no problems such as damaged edge or texture details, boundary artifacts and poor definition of the denoised image. The proposed method not only improves the image denoising ability, but also improves the denoising productivity to some extent. In the related work, we have made a comprehensive discussion on the current mainstream image denoising methods based on traditional machine learning and deep learning, which to some extent makes up for the lack of such reviews. In addition, the existing problems and future development direction of image denoising are discussed in detail.

show abstract

“…Besides, it can also be used to detect tissue changes in stroke and epilepsy. MRS data is acquired as a series of spectra, each representing the concentration of a different metabolite [2], [3]. The spectra are typically acquired in a 3D volume, with each voxel (3D pixel) in the volume containing a spectrum and can provide estimates of GABA levels in the brain, however, it has limitations and may not always be precise or accurate.…”

Section: Introductionmentioning

confidence: 99%

Magnetic Resonance Spectroscopy (MRS) Reconstruction Using Style Transfer Deep Depth wise Framework

Qayyum,

Niederer,

Khan

et al. 2024

Proceedings of International Conference on Artificial Life and Robotics

View full text Add to dashboard Cite

The human brain is a complex and heterogeneous organ composed of distinct compartments such as cerebral cortex, the cerebellum, the brainstem, and the subcortical regions. To analyze the chemical composition of tissues in brain, in vivo magnetic resonance spectroscopy allows non-invasive measurements of neurochemicals in either single voxel or multiple voxels. The reconstruction spectra using 1/3rd of original data than current Edited-MRS scans will not only result in four times faster edited-MRS scans but also extensively reduction in radiations. In this work, we present a deep depth-wise channel attention module (DCAM) based fine-tuned network for magnetic resonance spectroscopy image reconstruction. Besides, we have used channel-wise convolutions and average pooling without dimensionality reduction. We have trained the initial network from scratch on track-1 simulated dataset, however due to the limited dataset, we finetune the network on track-2 and track-3. Experiments are conducted on Edited-MRS-Rec-Challenge dataset1 that showed significantly better performance.

show abstract

Learning domain transfer for unsupervised magnetic resonance imaging restoration and edge enhancement

Cited by 10 publications

References 25 publications

Multi‐scale GAN with residual image learning for removing heterogeneous blur

Multi‐scale GAN with residual image learning for removing heterogeneous blur

A Novel Gray Image Denoising Method Using Convolutional Neural Network

Magnetic Resonance Spectroscopy (MRS) Reconstruction Using Style Transfer Deep Depth wise Framework

Contact Info

Product

Resources

About