A5100324487- Medical Imaging and Analysis
- Radiomics and Machine Learning in Medical Imaging
- Medical Image Segmentation Techniques
- Advanced Neural Network Applications
- Retinal Imaging and Analysis
- Glaucoma and retinal disorders
- Domain Adaptation and Few-Shot Learning
- Speech and Audio Processing
- Multimodal Machine Learning Applications
- Artificial Intelligence in Healthcare and Education
- Music and Audio Processing
- Cancer-related molecular mechanisms research
- AI in cancer detection
- Corneal surgery and disorders
- Brain Tumor Detection and Classification
- Genomics and Phylogenetic Studies
- Retinal and Optic Conditions
- Fluid Dynamics and Turbulent Flows
- Glioma Diagnosis and Treatment
- Phonocardiography and Auscultation Techniques
- Identification and Quantification in Food
- Aerodynamics and Acoustics in Jet Flows
- Plasma and Flow Control in Aerodynamics
- Machine Learning in Bioinformatics
- Lipoproteins and Cardiovascular Health
Northwestern Polytechnical University
2015-2024
Sichuan Agricultural University
2024
Ministry of Agriculture and Rural Affairs
2024
University of South China
2024
Chinese Academy of Medical Sciences & Peking Union Medical College
2024
University of Michigan–Ann Arbor
2023
Politecnico di Milano
2023
University of Shanghai for Science and Technology
2023
Guangdong University of Technology
2022
Guangdong Ocean University
2021-2022
Vehicle trajectories are one of the most important data in location-based services. The quality directly affects However, real applications, trajectory not always sampled densely. In this paper, we study problem recovering entire route between two distant consecutive locations a trajectory. Most existing works solve without using those informative historical or it an empirical way. We claim that data-driven and probabilistic approach is actually more suitable as long sparsity can be well...
Glaucoma is one of the leading causes irreversible blindness. Segmentation optic disc (OD) and cup (OC) on fundus images a crucial step in glaucoma screening. Although many deep learning models have been constructed for this task, it remains challenging to train an OD/OC segmentation model that could be deployed successfully different healthcare centers. The difficulties mainly comes from domain shift issue, i.e., collected at these centers usually vary greatly tone, contrast, brightness. To...
Self-supervised learning (SSL) has long had great success in advancing the field of annotation-efficient learning. However, when applied to CT volume segmentation, most SSL methods suffer from two limitations, including rarely using information acquired by different imaging modalities and providing supervision only bottleneck encoder layer. To address both we design a pretext task align each 3D corresponding 2D generated X-ray image extend self-distillation deep self-distillation. Thus,...
While MLLMs perform well on perceptual tasks, they lack precise multimodal alignment, limiting performance. To address this challenge, we propose Vision Dynamic Embedding-Guided Pretraining (VDEP), a hybrid autoregressive training paradigm for MLLMs. Utilizing dynamic embeddings from the MLP following visual encoder, approach supervises image hidden states and integrates tokens into training. Existing primarily focused recovering information textual inputs, often neglecting effective...
Cross-domain joint segmentation of optic disc and cup on fundus images is essential, yet challenging, for effective glaucoma screening. Although many unsupervised domain adaptation (UDA) methods have been proposed, these can hardly achieve complete alignment, leading to suboptimal performance. In this paper, we propose a triple-level alignment (TriLA) model address issue by aligning the source target domains at input level, feature output level simultaneously. At learnable Fourier (LFDA)...
Coelomactra antiquata is an important aquatic economic shellfish with high medicinal value. However, because C. has no reference genome, a lot of molecular biology research cannot be carried out, so the analysis its transcripts step to study regulatory genes various substances in antiquata. In present study, we conducted first full-length transcriptome by using PacBio single-molecule real-time (SMRT) sequencing technology. The results identified total 39,209 unigenes average length 2,732 bp,...
The images and sounds that we perceive undergo subtle but geometrically consistent changes as rotate our heads. In this paper, use these cues to solve a problem call Sound Localization from Motion (SLfM): jointly estimating camera rotation localizing sound sources. We learn tasks solely through self-supervision. A visual model predicts pair of images, while an audio the direction sources binaural sounds. train models generate predictions agree with one another. At test time, can be deployed...
It is well-known that(see e.g. Proposition 1.3.17 in [10]) the normalized principal eigenfunction $ \phi_1 of single elliptic eigenvalue problem$ -d\Delta \phi -c(x)\phi = \lambda\phi $with Robin boundary condition converges to 0 as d\rightarrow 0^+ locally uniformly \{x\in\bar\Omega|\; c(x)<\max_{\bar{\Omega}} c(x)\} $. The method used [10] designed for equation and seems difficult be applied systems directly. In this paper, we extend conclusion above system by introducing a different approach.
Universal segmentation models offer significant potential in addressing a wide range of tasks by effectively leveraging discrete annotations. As the scope and modalities expands, it becomes increasingly important to generate strategically position task- modal-specific priors within universal model. However, existing often overlook correlations between different priors, optimal placement frequency these remain underexplored. In this paper, we introduce MedUniSeg, prompt-driven model designed...
LncRNA plays a significant role in regulating feed efficiency. This study aims to explore the key long non-coding RNAs, associated genes, and pathways pigs with extreme efficiencies.
Medical images often exhibit distribution shifts due to variations in imaging protocols and scanners across different medical centers. Domain Generalization (DG) methods aim train models on source domains that can generalize unseen target domains. Recently, the segment anything model (SAM) has demonstrated strong generalization capabilities its prompt-based design, gained significant attention image segmentation tasks. Existing SAM-based approaches attempt address need for manual prompts by...
Parameter-efficient fine-tuning (PEFT) techniques have emerged to address issues of overfitting and high computational costs associated with fully in the paradigm self-supervised learning. Mainstream methods based on PEFT involve adding a few trainable parameters while keeping pre-trained backbone fixed. These achieve comparative, often superior, performance fine-tuning, demonstrating powerful representation ability backbone. Despite its success, these typically ignore initialization new...
How can we test AI performance? This question seems trivial, but it isn't. Standard benchmarks often have problems such as in-distribution and small-size sets, oversimplified metrics, unfair comparisons, short-term outcome pressure. As a consequence, good performance on standard does not guarantee success in real-world scenarios. To address these problems, present Touchstone, large-scale collaborative segmentation benchmark of 9 types abdominal organs. is based 5,195 training CT scans from...
Domain generalization (DG) aims to enhance the ability of models trained on source domains generalize effectively unseen domains. Recently, Sharpness-Aware Minimization (SAM) has shown promise in this area by reducing sharpness loss landscape obtain more generalized models. However, SAM and its variants sometimes fail guide model toward a flat minimum, their training processes exhibit limitations, hindering further improvements generalization. In paper, we first propose an improved process...