A5101759770- Medical Imaging Techniques and Applications
- Advanced X-ray and CT Imaging
- Neural Networks and Applications
- Radiomics and Machine Learning in Medical Imaging
- Advanced Chemical Sensor Technologies
- Identification and Quantification in Food
- Seismic Imaging and Inversion Techniques
- Biosensors and Analytical Detection
- Radiation Dose and Imaging
- Machine Learning and Data Classification
- Advanced Neural Network Applications
- Smart Agriculture and AI
- Greenhouse Technology and Climate Control
- Domain Adaptation and Few-Shot Learning
- Advanced Image Fusion Techniques
- Innovation in Digital Healthcare Systems
- Anomaly Detection Techniques and Applications
- Smart Grid Security and Resilience
- Image and Signal Denoising Methods
- Seismology and Earthquake Studies
- Time Series Analysis and Forecasting
- Cardiac Imaging and Diagnostics
- 3D Shape Modeling and Analysis
- Leaf Properties and Growth Measurement
- Human Pose and Action Recognition
University of Massachusetts Lowell
2021-2025
Shanghai Electric (China)
2024
University of Florida
2022
Tianjin University
2019
Beijing Normal University
2017
Low-dose computed tomography (LDCT) denoising is an important problem in CT research. Compared to the normal dose (NDCT), LDCT images are subjected severe noise and artifacts. Recently many studies, vision transformers have shown superior feature representation ability over convolutional neural networks (CNNs). However, unlike CNNs, potential of was little explored so far. To fill this gap, we propose a Convolution-free Token2Token Dilated Vision Transformer for low-dose denoising. The...
Abstract Objective: X-ray photon-counting detectors (PCDs) have recently gained popularity due to their capabilities in energy discrimination power, noise suppression, and resolution refinement. The latest extremity computed tomography (PCCT) scanner leverages these advantages for tissue characterization, material decomposition, beam hardening correction, metal artifact reduction. However, technical challenges such as charge splitting pulse pileup can distort the spectrum compromise image...
Low-dose computed tomography (LDCT) reduces the X-ray radiation but compromises image quality with more noises and artifacts. A plethora of transformer models have been developed recently to improve LDCT quality. However, success a model relies on large amount paired noisy clean data, which is often unavailable in clinical applications. In computer vision natural language processing fields, masked autoencoders (MAE) proposed as an effective label-free self-pretraining method for...
Due to the potential harmful risks caused by excessive X-ray radiation dose, low-dose has emerged as one of major rules computed tomography (CT) for clinical applications. However, reducing dose severely leads degraded reconstructed CT image quality with noises and artifacts, it may seriously hamper diagnosis. Currently, model-driven deep learning (DL)-based dual-domain expert knowledge can reconstruct images from projections acquired in non-ideal conditions, they have been demonstrated...
In established network architectures, shortcut connections are often used to take the outputs of earlier layers as additional inputs later layers. Despite extraordinary effectiveness shortcuts, there remain open questions on mechanism and characteristics. For example, why shortcuts powerful? Why do generalize well? this article, we investigate expressivity generalizability a novel sparse topology. First, demonstrate that topology can empower one-neuron-wide deep approximate any univariate...
Abstract Due to the complex flow structure and non-uniform phase distribution in vertical upward gas-liquid two-phase flow, an eight-electrode rotating electric field conductance sensor is used obtain multi-channel signals. The patterns of are classified according images obtained from a high-speed camera. Then, we employ multivariate weighted multi-scale permutation entropy (MWMPE) detect instability pattern transition flow. Afterwards, compare results MWMPE with those single-channel...
Cardiac computed tomography (CT) is widely used for diagnosis of cardiovascular disease, the leading cause morbidity and mortality in world. Diagnostic performance depends strongly on temporal resolution CT images. To image beating heart, one can reduce scanning time by acquiring limited-angle projections. However, this leads to increased noise limited-angle-related artifacts. The goal paper reconstruct high quality cardiac images from
The distributed energy power system needs to provide sufficient and flexible computing on demand meet the increasing digitization intelligence requirements of smart grid. However, current distribution loads in is unbalanced, with data center continuously increasing, while there a large amount idle at edge. Meanwhile, are number real-time tasks system, which have strict execution deadlines require reasonable scheduling multi-level heterogeneous demands. Based aforementioned background issues,...
Low-dose computed tomography (LDCT) offers reduced X-ray radiation exposure but at the cost of compromised image quality, characterized by increased noise and artifacts. Recently, transformer models emerged as a promising avenue to enhance LDCT quality. However, success such relies on large amount paired noisy clean images, which are often scarce in clinical settings. In computer vision natural language processing, masked autoencoders (MAE) have been recognized powerful self-pretraining...
Cardiac computed tomography (CT) has emerged as a major imaging modality for the diagnosis and monitoring of cardiovascular diseases. High temporal resolution is essential to ensure diagnostic accuracy. Limited-angle data acquisition can reduce scan time improve resolution, but typically leads severe image degradation motivates improved reconstruction techniques. In this paper, we propose novel physics-informed score-based diffusion model (PSDM) limited-angle cardiac CT. At sampling time,...
Enzymatic browning is a major quality defect of packaged “ready-to-eat” fresh-cut lettuce salads. While there have been many research and breeding efforts to counter this problem, progress hindered by the lack technology identify quantify rapidly, objectively, reliably. Here, we report deep learning model for prediction. To best our knowledge, it first-of-its-kind on prediction using pretrained Siamese Quadratic Swin (SQ-Swin) transformer with several highlights. First, includes quadratic...
Enzymatic browning is a major quality defect of packaged "ready-to-eat" fresh-cut lettuce salads. While there have been many research and breeding efforts to counter this problem, progress hindered by the lack technology identify quantify rapidly, objectively, reliably. Here, we report deep learning model for score prediction. To best our knowledge, it first-of-its-kind on prediction using Siamese Quadratic Swin (SQ-Swin) transformer with several highlights. First, includes quadratic...
A neural network with the widely-used ReLU activation has been shown to partition sample space into many convex polytopes for prediction. However, parameterized way a and other machine learning models use imperfections, \textit{e}.\textit{g}., compromised interpretability complex models, inflexibility in decision boundary construction due generic character of model, risk being trapped shortcut solutions. In contrast, although non-parameterized can adorably avoid or downplay these issues,...
A neural network (NN) with the widely-used ReLU activation has been shown to partition sample space into many convex polytopes for prediction. However, parametric way a NN and other machine learning models use imperfections, e.g., compromised interpretability complex models, inflexibility in decision boundary construction due generic character of model, risk being trapped shortcut solutions. In contrast, although nonparameterized can adorably avoid or downplay these issues, they are usually...
Low-dose computed tomography (LDCT) offers reduced X-ray radiation exposure but at the cost of compromised image quality, characterized by increased noise and artifacts. Recently, transformer models emerged as a promising avenue to enhance LDCT quality. However, success such relies on large amount paired noisy clean images, which are often scarce in clinical settings. In fields computer vision natural language processing, masked autoencoders (MAE) have been recognized an effective label-free...