A5012016311- AI in cancer detection
- Privacy-Preserving Technologies in Data
- Radiomics and Machine Learning in Medical Imaging
- Advanced Neural Network Applications
- Medical Image Segmentation Techniques
- Cutaneous Melanoma Detection and Management
- Retinal Imaging and Analysis
- Digital Imaging for Blood Diseases
- Pancreatic and Hepatic Oncology Research
- Cognitive Computing and Networks
- Brain Tumor Detection and Classification
- Gastric Cancer Management and Outcomes
- Colorectal Cancer Screening and Detection
- COVID-19 diagnosis using AI
- Foreign Language Teaching Methods
- Advanced X-ray and CT Imaging
- Advanced Computational Techniques and Applications
- Surgical Simulation and Training
- Education, Innovation and Language Studies
- Domain Adaptation and Few-Shot Learning
- Systemic Sclerosis and Related Diseases
- Ideological and Political Education
- Artificial Intelligence in Healthcare and Education
Chinese University of Hong Kong
2023
Wuchang University of Technology
2023
Beihang University
2021-2022
Federated learning (FL) allows multiple medical institutions to collaboratively learn a global model without centralizing client data. It is difficult, if possible at all, for such commonly achieve optimal performance each individual client, due the heterogeneity of images from various scanners and patient demographics. This problem becomes even more significant when deploying unseen clients outside FL with distributions not presented during federated training. To optimize prediction...
Abstract Recent advancements in artificial intelligence have witnessed human-level performance; however, AI-enabled cognitive assistance for therapeutic procedures has not been fully explored nor pre-clinically validated. Here we propose AI-Endo, an intelligent surgical workflow recognition suit, endoscopic submucosal dissection (ESD). Our AI-Endo is trained on high-quality ESD cases from expert endoscopist, covering a decade time expansion and consisting of 201,026 labeled frames. The...
Test-time adaptation (TTA) has increasingly been an important topic to efficiently tackle the cross-domain distribution shift at test time for medical images from different institutions. Previous TTA methods have a common limitation of using fixed learning rate all samples. Such practice would be sub-optimal TTA, because data may arrive sequentially therefore scale change frequently. To address this problem, we propose novel dynamic adjustment method test-time adaptation, called DLTTA, which...
Uncertainty estimation plays an important role for future reliable deployment of deep segmentation models in safety-critical scenarios such as medical applications. However, existing methods uncertainty have been limited by the lack explicit guidance calibrating prediction risk and model confidence. In this work, we propose a novel fine-grained reward maximization (FGRM) framework, to address directly utilizing metric related function with reinforcement learning based tuning algorithm. This...
Federated learning (FL) has emerged with increasing popularity to collaborate distributed medical institutions for training deep networks. However, despite existing FL algorithms only allow the supervised setting, most hospitals in realistic usually cannot afford intricate data labeling due absence of budget or expertise. This paper studies a practical yet challenging problem, named \textit{Federated Semi-supervised Learning} (FSSL), which aims learn federated model by jointly utilizing from...
Federated learning (FL) allows multiple medical institutions to collaboratively learn a global model without centralizing client data. It is difficult, if possible at all, for such commonly achieve optimal performance each individual client, due the heterogeneity of images from various scanners and patient demographics. This problem becomes even more significant when deploying unseen clients outside FL with distributions not presented during federated training. To optimize prediction...
Test-time adaptation (TTA) has increasingly been an important topic to efficiently tackle the cross-domain distribution shift at test time for medical images from different institutions. Previous TTA methods have a common limitation of using fixed learning rate all samples. Such practice would be sub-optimal TTA, because data may arrive sequentially therefore scale change frequently. To address this problem, we propose novel dynamic adjustment method test-time adaptation, called DLTTA, which...
Despite recent progress on semi-supervised federated learning (FL) for medical image diagnosis, the problem of imbalanced class distributions among unlabeled clients is still unsolved real-world use. In this paper, we study a practical yet challenging FL (imFed-Semi), which allows all to have only data while server just has small amount labeled data. This imFed-Semi addressed by novel dynamic bank scheme, improves client training exploiting proportion information. scheme consists two parts,...