A5079154623- Image and Signal Denoising Methods
- Advanced Image Processing Techniques
- Sparse and Compressive Sensing Techniques
- Intracerebral and Subarachnoid Hemorrhage Research
- Advanced Image Fusion Techniques
- Advanced Neuroimaging Techniques and Applications
- Alzheimer's disease research and treatments
- Cerebrospinal fluid and hydrocephalus
- Dementia and Cognitive Impairment Research
- Advanced MRI Techniques and Applications
- Acute Ischemic Stroke Management
- Fetal and Pediatric Neurological Disorders
- Functional Brain Connectivity Studies
- Colorectal Cancer Screening and Detection
- Radiomics and Machine Learning in Medical Imaging
- Immunotherapy and Immune Responses
- Image Enhancement Techniques
- Angiogenesis and VEGF in Cancer
- Photoacoustic and Ultrasonic Imaging
- Cardiovascular Health and Disease Prevention
- Advanced Data Compression Techniques
- Image and Video Quality Assessment
- Cardiac Imaging and Diagnostics
- Random lasers and scattering media
- Parkinson's Disease Mechanisms and Treatments
University of Maryland, Baltimore
2024
University of Pennsylvania
2019-2024
Hansoh Pharma (China)
2024
The University of Texas Health Science Center at San Antonio
2021-2023
Institute for Neurodegenerative Disorders
2021-2023
Laboratoire d’Imagerie Biomédicale
2023
Zhengzhou University
2016-2019
Peking University
2014-2018
Local and non-local attention-based methods have been well studied in various image restoration tasks while leading to promising performance. However, most of the existing solely focus on one type attention mechanism (local or non-local). Furthermore, by exploiting self-similarity natural images, pixel-wise operations tend give rise deviations process characterizing long-range dependence due degeneration. To overcome these problems, this paper we propose a novel collaborative network...
Total variation (TV) regularization is widely used in image restoration to exploit the local smoothness of content. Essentially, TV model assumes a zero-mean Laplacian distribution for gradient at all pixels. However, real-world images are nonstationary general, and assumption pixel might be invalid, especially regions with strong edges or rich textures. This paper introduces nonlocal (NL) extension regularization, which models sparsity pixelwise content-adaptive distributions, reflecting...
This paper proposes a new image denoising approach using adaptive signal modeling and soft-thresholding. It improves the quality by regularizing all patches in based on distribution transform domain. Instead of global model for patches, it employs content models to address non-stationarity signals. The each patch is estimated individually can vary different bands locations. In particular, we allow individual have non-zero expectation. To estimate expectation variance parameters particular...
The recently invented retina-inspired spike camera has shown great potential for capturing dynamic scenes. Different from the conventional digital cameras that compact photoelectric information within exposure interval into a single snapshot, produces continuous stream to record light intensity variation process. For cameras, image reconstruction remains an important and challenging issue. To this end, paper develops spike-to-image neural network (Spk2ImgNet) reconstruct scene stream. In...
Importance Little is known about the associations of strict blood pressure (BP) control with microstructural changes in small vessel disease markers. Objective To investigate regional intensive vs standard BP biomarkers, such as white matter lesions (WMLs), fractional anisotropy (FA), mean diffusivity (MD), and cerebral flow (CBF). Design, Setting, Participants The Systolic Blood Pressure Intervention Trial (SPRINT) a multicenter randomized clinical trial that compared systolic (SBP) (SBP...
Deep learning has been demonstrated effective in many neuroimaging applications. However, scenarios, the number of imaging sequences capturing information related to small vessel disease lesions is insufficient support data-driven techniques. Additionally, cohort-based studies may not always have optimal or essential for accurate lesion detection. Therefore, it necessary determine which are crucial precise This study introduces a deep framework detect enlarged perivascular spaces (ePVS) and...
This paper proposes a new image denoising algorithm based on adaptive signal modeling and regularization. It improves the quality of images by regularizing each patch using bandwise distribution in transform domain. Instead global model for all patches an image, it employs content-dependent models to address non-stationarity signals also diversity among different bands. The is adaptively estimated individually. varies from one location another In particular, we consider have non-zero...
Background: This study aims to determine the efficacy and safety profile of aumolertinib in real-word treatment setting for advanced non-small-cell lung cancer (NSCLC) patients harboring epidermal growth factor receptor (EGFR) mutations. Methods: We retrospectively analyzed clinical data 173 EGFR-mutated NSCLC who received at Henan Cancer Hospital from April 2020 December 2022. Progression-free survival (PFS) overall (OS) were evaluated using Kaplan–Meier curves, while a Cox regression model...
Abstract Lobar cerebral microbleeds (CMBs) and localized non-hemorrhage iron deposits in the basal ganglia have been associated with brain aging, vascular disease neurodegenerative disorders. Particularly, CMBs are small lesions require multiple neuroimaging modalities for accurate detection. Quantitative susceptibility mapping (QSM) derived from vivo magnetic resonance imaging (MRI) is necessary to differentiate between content mineralization. We set out develop a deep learning-based...
Neuroimaging heterogeneity in dementia has been examined using single modalities. We evaluated the associations of magnetic resonance imaging (MRI) atrophy and flortaucipir positron emission tomography (PET) clusters across Alzheimer's disease (AD) spectrum.
Importance Enlarged perivascular spaces (ePVSs) have been associated with cerebral small-vessel disease (cSVD). Although their etiology may differ based on brain location, study of ePVSs has limited to specific regions; therefore, risk factors and significance remain uncertain. Objective Toperform a whole-brain investigation in large community-based cohort. Design, Setting, Participants This cross-sectional analyzed data from the Atrial Fibrillation substudy population-based Multi-Ethnic...
Background: Neuroimaging bears the promise of providing new biomarkers that could refine diagnosis dementia. Still, obtaining pathology data required to validate relationship between neuroimaging markers and neurological changes is challenging. Existing repositories are focused on a single pathology, too small, or do not precisely match findings. Objective: The repository introduced in this work, South Texas Alzheimer’s Disease research center repository, was designed address these...
Conventional image and video communication systems are usually designed with the objective being to maximize fidelity of reconstructed images measured by mean square errors (MSE). It is well known that metric MSE may not reflect visual quality perceived human eyes. Recent advancements in assessment tell us structural similarity (SSIM), especially gradient similarity, reveals perceptual more reliably. Inspired this observation, paper proposes a new approach, which conveys information an...
In image restoration tasks, priors generally utilize correlation within contents to predict the latent signal. this paper, we propose jointly exploit both intra- and inter-patch of input image, so as further reduce uncertainty unknown signal, thus improve prediction image. The proposed scheme evolves from low-rank regularization for non-local highly-correlated contents. Since underlying cost function pursue minimal rank is hard solve, use non-convex smooth surrogates penalty. Two such are...
Although it has been known that the tumor microenvironment affects angiogenesis, precise mechanism remains unclear. In this study, we simulated of human esophageal squamous cell carcinoma (ESCC) by conditioned medium (TCM) to assess influence on normal endothelial cells (NECs). We found TCM-induced NECs showed enhanced angiogenic properties, such as migration, invasion and tube formation. Moreover, expressed (TECs) markers at higher levels, which indicated TCM probably promoted angiogenesis...
Image restoration techniques generally use intrinsic correlations of image contents to reduce the uncertainty unknown signal and estimate latent ground truth. Local non-local correlation are two major kinds utilized. They different sources reflecting connections between data, but such a difference is not taken into consideration in most existing schemes. Typically, sparse representation based works same data exploit both local shared regularization. This paper aims fully separately so that...
Spike camera is a kind of neuromorphic sensor that uses novel ``integrate-and-fire'' mechanism to generate continuous spike stream record the dynamic light intensity at extremely high temporal resolution. However, as trade-off for resolution, its spatial resolution limited, resulting in inferior reconstruction details. To address this issue, paper develops network (SpikeSR-Net) super-resolve high-resolution image sequence from low-resolution binary streams. SpikeSR-Net designed based on...
Arterial spin labeling (ASL) perfusion MRI is the only non-invasive imaging technique for quantifying regional cerebral blood flow (CBF), which a fundamental physiological variable. ASL has relatively low signal-to-noise-ratio (SNR). In this study, we proposed novel denoising method by simultaneously exploiting inter- and intra-receive channel data correlations. including have been routinely acquired with multichannel coils but current methods are designed coil-combined data. Indeed,...
G-Cast is a wireless visual communication scheme that conveys information via image gradient. It inspired by the characteristics of human vision systems and can provide improved perceptual quality. power efficient but bandwidth demanding, because gradient data have double size original image. This paper presents named CG-Cast for scalable transmission in bandwidth-limited scenarios. employs compressive-gradient-based representation to describe perceptually sensitive details reduce...
Total-variation (TV) regularization is widely adopted in image restoration problems to exploit the feature that natural images are smooth with small gradient values at most regions. Basic TV method assumes identical zero-mean Laplacian distribution for gradients all pixels. However, real-world images, statistics of may not be stationary, and assumption valid either a specific pixel. This paper presents non-local extension restoration, called Non-Local Gradient Sparsity Regularization (NGSR)....
In this study we propose a novel cerebral microbleed (CMB) detection technique which simultaneously utilizes distribution information in dual domains and shape obtained by Fourier descriptor, does not rely on large set of training data. Specifically, the domain modeling aims to examine image content both gradient voxel domain, while descriptor further characterize candidate region. A labeled data is used form dualdomain as well coefficients. Then probability region containing CMB estimated...
This paper combines an adaptive reconstruction based approach and a learning technique into effective scheme for single image super-resolution. Unlike conventional schemes that adopt pre-trained dictionaries to tell the relationship between high-resolution (HR) low-resolution (LR) observation, proposed method attempts learn joint distribution of highly-correlated patch couples from input itself instead external dataset, so learnt models are specially tailored current patches thus can better...
Prior knowledge plays an important role in image denoising tasks. This paper utilizes the data of input to adaptively model prior distribution. The proposed scheme is based on observation that, for a natural image, matrix consisted its vectorized non-local similar patches low rank. We use non-convex smooth surrogate low-rank regularization, and view optimization problem from empirical Bayesian perspective. In such framework, parameter-free distribution derived grouped contents. Experimental...