A5056596552- Image and Signal Denoising Methods
- Advanced Image Processing Techniques
- Advanced Image Fusion Techniques
- Medical Imaging Techniques and Applications
- Sparse and Compressive Sensing Techniques
- Advanced MRI Techniques and Applications
- Radiomics and Machine Learning in Medical Imaging
- Advanced X-ray and CT Imaging
- Medical Image Segmentation Techniques
- Image Processing Techniques and Applications
- Advanced Neuroimaging Techniques and Applications
- Neural Networks and Applications
- MRI in cancer diagnosis
- Blind Source Separation Techniques
- Advanced Photocatalysis Techniques
- Image Enhancement Techniques
- AI in cancer detection
- Structural Health Monitoring Techniques
- Underwater Acoustics Research
- Cutaneous Melanoma Detection and Management
- Cardiac Imaging and Diagnostics
- Cardiovascular Disease and Adiposity
- Numerical methods in inverse problems
- Advanced Vision and Imaging
- Advanced Adaptive Filtering Techniques
Inner Mongolia University
2016-2025
Changzhou University
2023-2024
Mongolian National University of Medical Sciences
2024
Chinese University of Hong Kong
2024
Shanghai Jiao Tong University
2012-2020
National University of Mongolia
2020
Université Européenne de Bretagne
2012
Université de Bretagne Sud
2012
The image nonlocal self-similarity (NSS) prior refers to the fact that a local patch often has many similar patches it across and been widely applied in recently proposed machining learning algorithms for processing. However, there is no theoretical analysis on its working principle literature. In this paper, we discover potential causality between NSS low-rank property of color images, which also available grey images. A new group based scheme learn explicit models natural numerical patched...
Abstract Skin lesion segmentation is a crucial step for skin analysis and subsequent treatment. However, it still challenging task due to the irregular fuzzy borders, diversity of lesions. In this article, we propose Triple‐UNet, an organic combination three UNet architectures with suitable modules, automatically segment To enhance target object region image, design interest enhancement module (ROIE) that uses predicted score map first UNet. The enhanced image features learned by help second...
Abstract Background Magnetic Resonance Imaging (MRI) and Positron Emission Tomography (PET) stand as pivotal diagnostic tools for brain disorders, offering the potential mutually enriching disease perspectives. However, costs associated with PET scans inherent radioactivity have limited widespread application of PET. Furthermore, it is noteworthy to highlight promising high‐field ultra‐high‐field neuroimaging in cognitive neuroscience research clinical practice. With enhancement MRI...
In this paper, a new denoising algorithm to deal with the additive white Gaussian noise model is described. Following nonlocal (NL) means approach, we propose an adaptive estimator based on weighted average of observations taken in neighborhood weights depending similarity local patches. The idea compute that best minimize upper bound pointwise $L_2$ risk. framework estimation, show “oracle” are optimal if consider triangular kernels instead commonly used kernel. Furthermore, way...
Inter-residue contacts in proteins dictate the topology of protein structures. They are crucial for folding and structural stability. Accurate prediction residue especially long-range is important to quality ab inito structure modeling since they can enforce strong restraints assembly.In this paper, we present a new Residue-Residue Contact predictor called R2C that combines machine learning-based correlated mutation analysis-based methods, together with two-dimensional Gaussian noise filter...
Abstract Objective . Positron Emission Tomography and Magnetic Resonance Imaging (PET-MRI) systems can obtain functional anatomical scans. But PET suffers from a low signal-to-noise ratio, while MRI are time-consuming. To address time-consuming, an effective strategy involves reducing k-space data collection, albeit at the cost of lowering image quality. This study aims to leverage inherent complementarity within PET-MRI enhance quality PET-MRI. Approach A novel joint reconstruction model,...
With the widespread application of convolutional neural networks (CNNs), traditional model based denoising algorithms are now outperformed. However, CNNs face two problems. First, they computationally demanding, which makes their deployment especially difficult for mobile terminals. Second, experimental evidence shows that often over-smooth regular textures present in images, contrast to non-local models. In this letter, we propose a solution both issues by combining nonlocal algorithm with...
Abstract Parameter selection is crucial to regularization-based image restoration methods. Generally speaking, a spatially fixed parameter for the regularization term does not perform well both edge and smooth areas. A larger reduces noise better in areas but blurs regions, while small sharpens causes residual noise. In this paper, an automated adaptive model, which combines harmonic total variation (TV) terms, proposed reconstruction from noisy blurred observation. The model detects edges...
We introduce an oracle filter for removing the Gaussian noise with weights depending on a similarity function. The usual Non-Local Means is obtained from this by substituting function estimator based patches. When sizes of search window are chosen appropriately, it shown that converges optimal rate. same convergence rate preserved when has suitable errors-in measurements. also provide statistical which at convenient Based our theorems, we propose some simple formulas choice parameters....
MRI and PET are important modalities can provide complementary information for the diagnosis of brain diseases because structural obtain functional brain. However, is usually missing. Especially, simultaneous imaging not achievable at ultrahigh field in current. Thus, synthetic using essential. In this paper, we as a guide by joint probability distribution diffusion model (JPDDM). Meanwhile, We utilized our fields.
This study aims to assess the consistency of various CT-FFR software, determine reliability current and measure relevant influence factors. The goal is build a solid foundation enhanced workflow technical principles that will ultimately improve accuracy measurements coronary blood flow reserve fractions. improvement critical for assessing level ischemia in patients with heart disease.
A new image denoising algorithm to deal with the additive Gaussian white noise model is given. Like non-local means method, filter based on weighted average of observations in a neighborhood, weights depending similarity local patches. But contrast filter, instead using fixed kernel, we propose choose by minimizing tight upper bound mean square error. This approach makes it possible define adapted function at hand, mimicking oracle filter. Under some regularity conditions target image, show...
This study describes an improved method for Poisson image denoising that is based on a state‐of‐the‐art approach known as non‐local principal component analysis (NLPCA). The new referred to PieceWise Principal Component Analysis (PWPCA). In PWPCA, the given first split into pieces, then NLPCA run each piece, and finally entire reconstituted by weighted combination of NLPCA‐processed pieces. Using standard test images with noise, authors show PWPCA restores more effectively than approaches....
Image denoising is one of the most important tasks in image processing. In this paper, we study methods by using similar patches which have low-rank covariance matrices to recover an underlying corrupted additive Gaussian noise. order enhance global patch-matching results, make use a mixture model with auxiliary determine different groups patches. The output BM3D. noisy version matrix formed each group from given image. Its can be estimated nuclear norm minimization, and resulting denoised...