A5101806571- Neonatal and fetal brain pathology
- Fetal and Pediatric Neurological Disorders
- Functional Brain Connectivity Studies
- Domain Adaptation and Few-Shot Learning
- Advanced Neuroimaging Techniques and Applications
- Advanced Neural Network Applications
- Generative Adversarial Networks and Image Synthesis
- Metaheuristic Optimization Algorithms Research
- Advanced Multi-Objective Optimization Algorithms
- EEG and Brain-Computer Interfaces
- Advanced Image Processing Techniques
- Image and Signal Denoising Methods
- Advanced MRI Techniques and Applications
- Machine Learning and ELM
- Extracellular vesicles in disease
- Museums and Cultural Heritage
- Surface Roughness and Optical Measurements
- Optimal Experimental Design Methods
- Neural dynamics and brain function
- Speech and Audio Processing
- Multimodal Machine Learning Applications
- Advanced Optical Imaging Technologies
- Software Engineering Techniques and Practices
- Interactive and Immersive Displays
- Image Processing Techniques and Applications
University of North Carolina at Chapel Hill
2020-2025
Xi'an Jiaotong University
2015-2024
Imaging Center
2023
Jiangnan University
2023
Southwest Jiaotong University
2021
Xiangtan University
2006-2011
Tsinghua University
2009-2010
University Town of Shenzhen
2010
State Administration of Cultural Heritage
2010
Institute of Information Engineering
2007
Converging evidence increasingly suggests that psychiatric disorders, such as major depressive disorder (MDD) and autism spectrum (ASD), are not unitary diseases, but rather heterogeneous syndromes involve diverse, co-occurring symptoms divergent responses to treatment. This clinical heterogeneity has hindered the progress of precision diagnosis treatment effectiveness in disorders. In this study, we propose BPI-GNN, a new interpretable graph neural network (GNN) framework for analyzing...
Spatiotemporal (four-dimensional) infant-dedicated brain atlases are essential for neuroimaging analysis of early dynamic development. However, due to the substantial technical challenges in acquisition and processing infant MR images, 4D densely covering development during infancy still scarce. Few existing ones generally have fuzzy tissue contrast low spatiotemporal resolution, leading degraded accuracy atlas-based normalization subsequent analyses. To address this issue, paper, we...
The hippocampal formation is implicated in a myriad of crucial functions, particularly centered around memory and emotion, with distinct subdivisions fulfilling specific roles. However, there no consensus on the spatial organization these subdivisions, given that functional connectivity gene expression-based parcellation along its longitudinal axis differs from histology-based medial-lateral axis. dynamic nonuniform surface expansion hippocampus during early development reflects underlying...
Precise segmentation of subcortical structures from infant brain magnetic resonance (MR) images plays an essential role in studying early structural and functional developmental patterns diagnosis related disorders. However, due to the dynamic appearance changes, low tissue contrast, tiny size MR images, is a challenging task. In this paper, we propose context-guided, attention-based, coarse-to-fine deep framework precisely segment structures. At coarse stage, aim directly predict signed...
The human brain development experiences a complex evolving cortical folding from smooth surface to convoluted ensemble of folds. Computational modeling has played an essential role in better understanding the process folding, but still leaves many questions be answered. A major challenge faced by computational models is how create massive developmental simulations with affordable sources complement neuroimaging data and provide reliable predictions for folding. In this study, we leveraged...
Abstract Brain subcortical structures are paramount in many cognitive functions and their aberrations during infancy predisposed to various neurodevelopmental neuropsychiatric disorders, making it highly essential characterize the early normative growth patterns. This study investigates volumetric development surface area expansion of six associations with Mullen scales learning by leveraging 513 high-resolution longitudinal MRI scans within first two postnatal years. Results show that (1)...
Longitudinal brain imaging atlases with densely sampled time-points and ancillary anatomical information are of fundamental importance in studying early developmental characteristics human non-human primate brains during infancy, which feature extremely dynamic appearance, shape size. However, for primates, highly valuable animal models understanding brains, the existing mainly developed based on adults or adolescents, denoting a notable lack temporally densely-sampled covering development....
Revealing early dynamic development of the normative cerebellar structures contributes to exploring cerebellum-related neurodevelopmental disorders. Here, leveraging infant-tailored image processing techniques, we studied volumetric developmental trajectories cerebellum and 27 sub-regions their relationships with behavioral scores based on 511 high-resolution structural MRI scans during first 800 postnatal days. The ratio entire intracranial volume increases rapidly at then peaks 13 months...
Through multiple levels of abstraction, deep learning takes advantage layers models to find the complicated structure and learn high level representations data. In recent years, has made great progress in object detection, speech recognition, many other domains. The robustness systems with architectures is however rarely studied needs further investigation. Especially, mean square error(MSE), which commonly used as optimization cost function learning, sensitive outliers(or impulsive noises)....
Robust motion correction of fetal brain MRI slices is crucial for 3D volume reconstruction. However, conventional methods can only handle a limited range motion. Hence, deep learning model based on geometric constraints proposed in order to predict the arbitrary standard anatomical space, which consists global estimation network and relative network. In particular, used estimate between two adjacent slices, exploited as constraint. Then, sharing features networks make learn more unique...
Robust motion correction of fetal brain MRI slices is crucial for volume reconstruction. However, conventional methods can only handle a limited range motion. Hence, deep learning model based on prior geometric constraints proposed to predict the 2D slices. It consists global and relative estimation network. Sharing features between two networks make learn more unique feature representations correction. Moreover, we present control point-based approach simulate complex trajectories. The...
Despite the remarkable code generation abilities of large language models (LLMs), they still face challenges in complex task handling. Robot development, a highly intricate field, inherently demands human involvement allocation and collaborative teamwork[1]. To enhance robot we propose an innovative automated collaboration framework inspired by real-world developers. This employs multiple LLMs distinct roles-analysts, programmers, testers. Analysts delve deep into user requirements, enabling...
Short throw interactive projection systems are widely used in education, training, commerce, and entertainment recent years. Different techniques have been developed. And among them, the infrared location technique is one of attracting methods, because advantages such as independent whiteboard low cost, so on. However, main defect that pen point easy to be blocked by user's hand. In this paper, we introduced our progress on indirect measurement a short system. Two LEDs fixed along pen's body...
In this paper, we describe a multi-objective particle swarm optimization algorithm (MOPSO) that incorporates the concept of enhanced ε -dominance, present new to update archive, archiving technique can help us maintain sequence well-spread solutions. A strategy and mutation operator are shown speed up convergence. To compare with state-of-art MOEAs on well-established suite test problems, our approach is simple constructed, results indicate it works effective has steady-state performance. It...
The design of a visual optical system for eye and vision examination using video-based chart is investigated in this paper. It produced virtual image larger than the original display much farther away from tester eye. diagonal length enlarged as designed to match resolution human provides wide observing area. also has large field view. comfortable long relief. results analysis are discussed