A5100406322- COVID-19 diagnosis using AI
- Radiomics and Machine Learning in Medical Imaging
- AI in cancer detection
- Anomaly Detection Techniques and Applications
- Advanced Neural Network Applications
- Medical Imaging Techniques and Applications
- Image Enhancement Techniques
- Medical Image Segmentation Techniques
- Handwritten Text Recognition Techniques
- Domain Adaptation and Few-Shot Learning
- Vehicle License Plate Recognition
- Traditional Chinese Medicine Studies
- Gait Recognition and Analysis
- Infrastructure Maintenance and Monitoring
- Human Pose and Action Recognition
- Industrial Vision Systems and Defect Detection
- Advanced X-ray and CT Imaging
- Colorectal Cancer Screening and Detection
- Machine Learning and ELM
- Artificial Intelligence in Healthcare and Education
- Neural Networks and Applications
- Data Stream Mining Techniques
- Advanced Algorithms and Applications
- Generative Adversarial Networks and Image Synthesis
- Advanced Image and Video Retrieval Techniques
Hokkaido University
2020-2025
Southeast University
2019-2024
Hokkaido University of Education
2024
Tulane University
2020
Louisiana State University Health Sciences Center New Orleans
2020
Southeast Louisiana Veterans Health Care System
2020
Fo Guang University
2020
Rensselaer Polytechnic Institute
2019
Neusoft (China)
2018
National Institutes of Health
2011
heart failure ◼ SARS virus severe acute respiratory syndrome T his Research Letter expands our previous report of 10 hearts 1 by adding an additional 12 (for a total 22 hearts) from deaths confirmed attributable to coronavirus disease 2019 (COVID-19) infection.We identify key gross and microscopic changes that challenge the notion typical myocarditis is present in 2 speculate on alternative mechanisms for cardiac injury should be investigated provide better understanding manifestations...
Dataset distillation is an effective technique for reducing the cost and complexity of model training while maintaining performance by compressing large datasets into smaller, more efficient versions. In this paper, we present a novel generative dataset method that can improve accuracy aligning prediction logits. Our approach integrates self-knowledge to achieve precise distribution matching between synthetic original data, thereby capturing overall structure relationships within data. To...
In few-shot action recognition (FSAR), long sub-sequences of video naturally express entire actions more effectively. However, the high computational complexity mainstream Transformer-based methods limits their application. Recent Mamba demonstrates efficiency in modeling sequences, but directly applying to FSAR overlooks importance local feature and alignment. Moreover, within same class accumulate intra-class variance, which adversely impacts performance. To solve these challenges, we...
Challenges drive the state-of-the-art of automated medical image analysis. The quantity public training data that they provide can limit performance their solutions. Public access to methodology for these solutions remains absent. This study implements Type Three (T3) challenge format, which allows on private and guarantees reusable methodologies. With T3, organizers train a codebase provided by participants sequestered data. T3 was implemented in STOIC2021 challenge, with goal predicting...
In this study, we propose a novel dataset distillation method based on parameter pruning. The proposed can synthesize more robust distilled datasets and improve performance by pruning difficult-to-match parameters during the process. Experimental results two benchmark show superiority of method.
The global outbreak of the Coronavirus 2019 (COVID-19) has overloaded worldwide healthcare systems. Computer-aided diagnosis for COVID-19 fast detection and patient triage is becoming critical. This paper proposes a novel self-knowledge distillation based self-supervised learning method from chest X-ray images. Our can use images on similarities their visual features learning. Experimental results show that our achieved an HM score 0.988, AUC 0.999, accuracy 0.957 largest open dataset.
This paper proposes a novel self-supervised learning method for better representations with small batch sizes. Many methods based on certain forms of the siamese network have emerged and received significant attention. However, these need to use large sizes learn good require heavy computational resources. We present new triplet combined triple-view loss improve performance representation Experimental results show that our can drastically outperform state-of-the-art several datasets in...
Traffic sign recognition is a complex and challenging yet popular problem that can assist drivers on the road reduce traffic accidents. Most existing methods for use convolutional neural networks (CNNs) achieve high accuracy. However, these first require large number of carefully crafted datasets training process. Moreover, since signs differ in each country there variety signs, need to be fine-tuned when recognizing new categories. To address issues, we propose matching method zero-shot...
Poor chip solder joints can severely affect the quality of finished printed circuit boards (PCBs). Due to diversity joint defects and scarcity anomaly data, it is a challenging task automatically accurately detect all types in production process real time. To address this issue, we propose flexible framework based on contrastive self-supervised learning (CSSL). In framework, first design several special data augmentation approaches generate abundant synthetic, not good (sNG) from normal...
There is an urgent need to find effective and accurate method for triaging coronavirus disease 2019 (COVID-19) patients from millions or billions of people. Therefore, this study aimed develop a novel deep-learning approach COVID-19 triage based on chest computed tomography (CT) images, including normal, pneumonia, cases.A total 2,809 CT scans (1,105 COVID-19, 854 850 non-3COVID-19 pneumonia cases) were acquired classified into the training set (n = 2,329) test 480). A U-net-based...
The synthesis of medical images from one modality to another is an intensity transformation between two acquired different devices, such as Magnetic Resonance (MR)image Computed Tomography (CT)image, or MR T1 weighted (T1W)image T2 (T2W)or proton density (PDW)image. based synthetic CT very useful for some clinical cases, PET attenuation correction PET/MR, MR/CT registration etc. In this paper, we propose a novel method on fully convolutional networks (FCN)to generate image image. We adopt...
In our digitally driven society, advances in software and hardware to capture video data allow extensive gathering analysis of large datasets. This has stimulated interest extracting information from data, such as buildings urban streets, enhance understanding the environment. Urban essential parts cities, carry valuable relevant daily life. Extracting features these elements integrating them with technologies VR AR can contribute more intelligent personalized public services. Despite its...
A novel cross-view self-supervised learning (CVSSL) method via momentum statistics in batch normalization is presented this paper. The problem of accuracy degradation small-batch cases currently common learning. Our introduces the loss and to solve cases. Experimental results show that our can drastically outperform state-of-the-art on STL-10 dataset.
Manually annotating gastric X-ray images for gastritis detection is time-consuming and expensive because it typically requires expert knowledge. In this paper, we propose a novel self-supervised learning method with scarce annotations. Our introduces triplet networks triple-view loss to solve the insufficient available annotations in detection. Experimental results show that our can outperform several state-of-the-art methods