A5053007087- Functional Brain Connectivity Studies
- Advanced Neuroimaging Techniques and Applications
- AI in cancer detection
- Dementia and Cognitive Impairment Research
- Hydraulic Fracturing and Reservoir Analysis
- Neural dynamics and brain function
- Radiomics and Machine Learning in Medical Imaging
- Geotechnical and Geomechanical Engineering
- COVID-19 diagnosis using AI
- Alzheimer's disease research and treatments
- Seismic Performance and Analysis
- Coal Combustion and Slurry Processing
- Neurological Disease Mechanisms and Treatments
- Advanced MRI Techniques and Applications
- Magnesium Alloys: Properties and Applications
- Energetic Materials and Combustion
- Seismic Imaging and Inversion Techniques
- Aluminum Alloy Microstructure Properties
- Embodied and Extended Cognition
- EEG and Brain-Computer Interfaces
- Diamond and Carbon-based Materials Research
- Innovative Human-Technology Interaction
- Granular flow and fluidized beds
- Neonatal and fetal brain pathology
- High-pressure geophysics and materials
Wuhan Polytechnic University
2025
The University of Texas at Arlington
2021-2024
University of Michigan
2022-2024
Southwest Petroleum University
2024
Arizona State University
2020-2023
Guangxi Medical University
2023
Lawrence Livermore National Laboratory
2023
China University of Petroleum, Beijing
2019-2022
Deep models are powerful in capturing the complex and non-linear relationship buried brain imaging data. However, huge number of parameters deep can easily overfit given limited data samples. In this work, we proposed a cross-domain transfer learning method to solve insufficient problem domain by leveraging knowledge learned natural image domain. Specifically, employed ViT as backbone firstly pretrained it using ImageNet-21K dataset then transferred dataset. A slice-wise convolution...
Technologically-supported social interaction has gained significant attention within the Human-Computer Interaction community, particularly for facilitating remote connections. However, less emphasis been placed on co-located situations and multi-user scenarios where participants have unfamiliar relationships. We propose an augmented multimodal media method interaction. Our study identifies three design principles designing technologically-supported responsive environments: (1) over time,...
To solve the missed and wrong detection problems of object model in identifying soybean companion weeds, this paper proposes an enhanced multi-scale channel feature based on RT-DETR (EMCF-RTDETR). First, we designed a lightweight hybrid-channel extraction backbone network, which consists CGF-Block module FasterNet-Block working together, aiming to reduce amount computation number parameters while improving efficiency extraction. Second, constructed EA-AIFI module. This enhances detailed...
Evolution of nitrogen under shock compression up to 100 GPa is revisited via molecular dynamics simulations using a machine-learned interatomic potential. The model shown be capable recovering the structure, dynamics, speciation, and kinetics in hot compressed liquid predicted by first-principles as well measured principal Hugoniot double experimental data, albeit without cooling. Our results indicate that purely dissociation description chemistry provides an incomplete picture short...
We present a new parameterization of the ChIMES physics informed machine- learned interatomic model for simulating carbon under conditions ranging from 300 K and 0 GPa to 10,000 100 GPa, along with multi-fidelity active learning strategy. The resulting shows significant improvement in accuracy temperature/pressure transferability relative original developed 2017, can serve as foundation future transfer-learned parameter sets. Model applications melting point prediction, shockwave-driven con-...
We present a new parameterization of the ChIMES physics informed machine- learned interatomic model for simulating carbon under conditions ranging from 300 K and 0 GPa to 10,000 100 GPa, along with multi-fidelity active learning strategy. The resulting shows significant improvement in accuracy temperature/pressure transferability relative original developed 2017, can serve as foundation future transfer-learned parameter sets. Model applications melting point prediction, shockwave-driven con-...
Designing more efficient, reliable, and explainable neural network architectures is critical to studies that are based on artificial intelligence (AI) techniques. Previous studies, by post-hoc analysis, have found the best-performing ANNs surprisingly resemble biological networks (BNN), which indicates BNNs may share some common principles achieve optimal performance in either machine learning or cognitive/behavior tasks. Inspired this phenomenon, we proactively instill organizational of...
In this paper, we present a large-scale evaluation probing GPT-4V's capabilities and limitations for biomedical image analysis. GPT-4V represents breakthrough in artificial general intelligence (AGI) computer vision, with applications the domain. We assess performance across 16 medical imaging categories, including radiology, oncology, ophthalmology, pathology, more. Tasks include modality recognition, anatomy localization, disease diagnosis, report generation, lesion detection. The...
Abstract In this prospective paper, we first review the existing simulation tools to simulate microgalvanic corrosion during free immersion. Then, describe a recently developed application that employs PRISMS-PF, an open-source, high-performance phase-field modeling framework. The model employed in accounts for electrochemical reaction at metal/electrolyte interface and ionic migration electrolyte determine evolution of front. We present implementation details discuss its features such as...
How to identify and characterize functional brain networks (BN) is fundamental gain system-level insights into the mechanisms of organizational architecture. Current magnetic resonance (fMRI) analysis highly relies on prior knowledge specific patterns in either spatial (e.g., resting-state network) or temporal task stimulus) domain. In addition, most approaches aim find group-wise common networks, individual-specific have been rarely studied. this work, we propose a novel Twin-Transformers...
We present a new parameterization of the ChIMES physics informed machine- learned interatomic model for simulating carbon under conditions ranging from 300 K and 0 GPa to 10,000 100 GPa, along with multi-fidelity active learning strategy. The resulting shows significant improvement in accuracy temperature/pressure transferability relative original developed 2017, can serve as foundation future transfer-learned parameter sets. Model applications melting point prediction, shockwave-driven con-...
The human cerebral cortex is highly convoluted into convex gyri and concave sulci. It has been demonstrated that sulci are significantly different in their anatomy, connectivity, function, besides exhibiting opposite shape patterns, long-distance axonal fibers connected to much denser than those sulci, neural signals on more complex low-frequency while high-frequency. Although accumulating evidence shows significant differences between primary roles brain function have not elucidated yet. To...
With the popularity of deep neural networks (DNNs), model interpretability is becoming a critical concern. Many approaches have been developed to tackle problem through post-hoc analysis, such as explaining how predictions are made or understanding meaning neurons in middle layers. Nevertheless, these methods can only discover patterns rules that naturally exist models. In this work, rather than relying on schemes, we proactively instill knowledge alter representation human-understandable...
As a progressive neurodegenerative disorder, the pathological changes of Alzheimer's disease (AD) might begin as much two decades before manifestation clinical symptoms. Since nature irreversible pathology AD, early diagnosis provides more tractable way for intervention and treatment. Therefore, numerous approaches have been developed diagnostic purposes. Although several important biomarkers established, most existing methods show limitations in describing continuum AD progression. However,...
We studied the association between socioeconomic status (SES), tooth loss, and oral health-related quality of life (OHRQoL) in an adult cohort western China. As inequalities health are often neglected promotion. we aimed to verify impact SES on loss OHRQoL.
As reversing the pathology of Alzheimer's disease (AD) is impossible, diagnosis mild cognitive impairment (MCI), which considered as precursor AD, has become a more tractable goal. Because both brain structural and functional alterations have been observed in MCI patients, many multimodal fusion approaches proposed to classify from normal controls (NC) clinical studies. Given complex relationships between structure function, deep learning based models can be helpful revealing potential...
Mild cognitive impairment (MCI) is a high-risk dementia condition which progresses to probable Alzheimer's disease (AD) at approximately 10% 15% per year. Characterization of group-level differences between two subtypes MCI – stable (sMCI) and progressive (pMCI) the key step understand mechanisms progression enable possible delay transition from AD. Functional connectivity (FC) considered as promising way study since may show alterations even in preclinical stages provide substrates for AD...
Recently, a novel cortical folding pattern known as the 3-hinge gyrus (3HG) has been identified. 3HGs are defined convergence of gyri coming from three distinct directions on gyral crests. In contrast to regions, at finer scale and they widely exist across different individuals, representing both commonalities individualities patterns. It is important note that identified in individual spaces, lacking natural cross-subject correspondences. To address this issue, we have developed...
Mild cognitive impairment (MCI) is recognized as a precursor to Alzheimer's disease (AD), progressive and irreversible neurodegenerative disorder of the brain. The neurodegeneration brain connectivity networks plays pivotal role in development progression MCI. Traditionally, are generated using coarse-grained regions, where regions serve nodes their functional or structural connections used edges. Recently, novel finer scale folding patterns named 3-hinge gyrus (3HG) was identified, which...