A5100658356- Radiomics and Machine Learning in Medical Imaging
- AI in cancer detection
- COVID-19 diagnosis using AI
- Advanced Neural Network Applications
- Lung Cancer Diagnosis and Treatment
- Domain Adaptation and Few-Shot Learning
- Medical Imaging and Analysis
- Respiratory Support and Mechanisms
- Multimodal Machine Learning Applications
- Topic Modeling
- Anomaly Detection Techniques and Applications
- Cutaneous Melanoma Detection and Management
- Advanced Graph Neural Networks
- Medical Image Segmentation Techniques
- Nonmelanoma Skin Cancer Studies
- Neonatal Respiratory Health Research
- Image Retrieval and Classification Techniques
- Advanced X-ray and CT Imaging
- Brain Tumor Detection and Classification
- Advanced Image and Video Retrieval Techniques
- Digital Imaging for Blood Diseases
- Advanced Clustering Algorithms Research
- Microbial Natural Products and Biosynthesis
- Cardiac Imaging and Diagnostics
- Medical Imaging Techniques and Applications
PLA Information Engineering University
2019-2025
Xiamen University
2025
Zhejiang University of Science and Technology
2024
Zhejiang University
2024
Alibaba Group (Cayman Islands)
2023-2024
Peking University
2024
Kunming Medical University
2024
Peking University Cancer Hospital
2024
Jilin University of Finance and Economics
2024
Alibaba Group (China)
2023
Automated skin lesion classification in dermoscopy images is an essential way to improve the diagnostic performance and reduce melanoma deaths. Although deep convolutional neural networks (DCNNs) have made dramatic breakthroughs many image tasks, accurate of lesions remains challenging due insufficiency training data, inter-class similarity, intra-class variation, lack ability focus on semantically meaningful parts. To address these issues, we propose attention residual learning network...
The accurate identification of malignant lung nodules on chest CT is critical for the early detection cancer, which also offers patients best chance cure. Deep learning methods have recently been successfully introduced to computer vision problems, although substantial challenges remain in due lack large training data sets. In this paper, we propose a multi-view knowledge-based collaborative (MV-KBC) deep model separate from benign using limited data. Our learns 3-D nodule characteristics by...
Clusters of viral pneumonia occurrences over a short period may be harbinger an outbreak or pandemic. Rapid and accurate detection using chest X-rays can significant value for large-scale screening epidemic prevention, particularly when other more sophisticated imaging modalities are not readily accessible. However, the emergence novel mutated viruses causes substantial dataset shift, which greatly limit performance classification-based approaches. In this paper, we formulate task...
Automated skin lesion segmentation and classification are two most essential related tasks in the computer-aided diagnosis of cancer. Despite their prevalence, deep learning models usually designed for only one task, ignoring potential benefits jointly performing both tasks. In this paper, we propose mutual bootstrapping convolutional neural networks (MB-DCNN) model simultaneous classification. This consists a coarse network (coarse-SN), mask-guided (mask-CN), an enhanced (enhanced-SN). On...
Due to the intensive cost of labor and expertise in annotating 3D medical images at a voxel level, most benchmark datasets are equipped with annotations only one type organs and/or tumors, resulting so-called partially labeling issue. To address this issue, we propose dynamic on-demand network (DoDNet) that learns segment multiple tumors on labeled datasets. DoD-Net consists shared encoder-decoder architecture, task encoding module, controller for filter generation, single but segmentation...
Cluster of viral pneumonia occurrences during a short period time may be harbinger an outbreak or pandemic, like SARS, MERS, and recent COVID-19. Rapid accurate detection using chest X-ray can significantly useful in large-scale screening epidemic prevention, particularly when other imaging modalities are less available. Viral often have diverse causes exhibit notably different visual appearances on images. The evolution viruses the emergence novel mutated further result substantial dataset...
The classification of medical images and illustrations from the biomedical literature is important for automated review, retrieval, mining. Although deep learning effective large-scale image classification, it may not be optimal choice this task as there only a small training dataset. We propose combined handcrafted visual feature (CDHVF) based algorithm that uses features learned by three fine-tuned pretrained convolutional neural networks (DCNNs) two descriptors in joint approach....
Automated segmentation of liver tumors in contrast-enhanced abdominal computed tomography (CT) scans is essential assisting medical professionals to evaluate tumor development and make fast therapeutic schedule. Although deep convolutional neural networks (DCNNs) have contributed many breakthroughs image segmentation, this task remains challenging, since 2D DCNNs are incapable exploring the inter-slice information 3D too complex be trained with available small dataset. In paper, we propose...
Medical image segmentation is an essential task in computer-aided diagnosis. Despite their prevalence and success, deep convolutional neural networks (DCNNs) still need to be improved produce accurate robust enough results for clinical use. In this paper, we propose a novel generic framework called Segmentation-Emendation-reSegmentation-Verification (SESV) improve the accuracy of existing DCNNs medical segmentation, instead designing more model. Our idea predict errors produced by model then...
The domain gap caused mainly by variable medical image quality renders a major obstacle on the path between training segmentation model in lab and applying trained to unseen clinical data. To address this issue, generalization methods have been proposed, which however usually use static convolutions are less flexible. In paper, we propose multi-source based content adaptive convolution (DCAC) for of images across different modalities. Specifically, design (DAC) module (CAC) incorporate both...
Analyzing the rich information behind heterogeneous networks through network representation learning methods is signifcant for many application tasks such as link prediction, node classifcation and similarity research. As evolve over times, interactions among nodes in make exhibit dynamic characteristics. However, almost all existing focus on static which ignore In this paper, we propose a novel approach DHNE to learn representations of networks. The key idea our construct comprehensive...
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,...