A5027122944- Advanced Neural Network Applications
- AI in cancer detection
- Radiomics and Machine Learning in Medical Imaging
- Liver Disease Diagnosis and Treatment
- Advanced MRI Techniques and Applications
- COVID-19 diagnosis using AI
- Cardiac Valve Diseases and Treatments
- Explainable Artificial Intelligence (XAI)
- Artificial Intelligence in Healthcare
- Domain Adaptation and Few-Shot Learning
- Medical Image Segmentation Techniques
- Cardiac Structural Anomalies and Repair
- Artificial Intelligence in Healthcare and Education
- Medical Imaging Techniques and Applications
- Adversarial Robustness in Machine Learning
- Algorithms and Data Compression
- Privacy-Preserving Technologies in Data
- Biomedical and Engineering Education
- Functional Brain Connectivity Studies
- Machine Learning and Algorithms
- Machine Learning in Bioinformatics
- Machine Learning and Data Classification
- Cardiovascular Function and Risk Factors
- MRI in cancer diagnosis
Fudan University
2022-2025
Huazhong University of Science and Technology
2017
In recent years, several deep learning models have been proposed to accurately quantify and diagnose cardiac pathologies. These automated tools heavily rely on the accurate segmentation of structures in MRI images. However, right ventricle is challenging due its highly complex shape ill-defined borders. Hence, there a need for new methods handle such structure's geometrical textural complexities, notably presence pathologies as Dilated Right Ventricle, Tricuspid Regurgitation,...
Large training datasets are important for deep learning-based methods. For medical image segmentation, it could be however difficult to obtain large number of labeled images solely from one center. Distributed learning, such as swarm has the potential use multi-center data without breaching privacy. However, distributions across centers can vary a lot due diverse imaging protocols and vendors (known feature skew). Also, regions interest segmented different, leading inhomogeneous label...
Accurate staging of liver fibrosis from magnetic resonance imaging (MRI) is crucial in clinical practice. While conventional methods often focus on a specific sub-region, multi-view learning captures more information by analyzing multiple patches simultaneously. However, previous approaches could not typically calculate uncertainty nature, and they generally integrate features different views black-box fashion, hence compromising reliability as well interpretability the resulting models. In...
Due to the high stakes in medical decision-making, there is a compelling demand for interpretable deep learning methods image analysis. Concept Bottleneck Models (CBM) have emerged as an active framework incorporating human-interpretable concepts into decision-making. However, their concept predictions may lack reliability when applied clinical diagnosis, impeding explanations' quality. To address this, we propose evidential Embedding Model (evi-CEM), which employs model uncertainty....
Cardiac segmentation is in great demand for clinical practice. Due to the enormous labor of manual delineation, unsupervised desired. The ill-posed optimization problem this task inherently challenging, requiring well-designed constraints. In work, we propose an framework multi-class with both intensity and shape Firstly, extend a conventional non-convex energy function as constraint implement it U-Net. For constraint, synthetic images are generated from anatomical labels via image-to-image...
Staging of liver fibrosis is important in the diagnosis and treatment planning patients suffering from diseases. Current deep learning-based methods using abdominal magnetic resonance imaging (MRI) usually take a sub-region as an input, which nevertheless could miss critical information. To explore richer representations, we formulate this task multi-view learning problem employ multiple sub-regions liver. Previously, features or predictions are combined implicit manner, uncertainty-aware...
Phosphorylated proteomics is a discipline that studies all phosphorylated proteins involved in the life process and plays an important role proteomics.With advancement of technology, there have been many experimental methods for protein phosphorylation bioinformatics-based computing strategies.The application data-driven machine learning phosphate-catalyzed research has matured become mainstream method this field.This paper mainly summarizes main principles learning, including Bayesian...
Distributed learning has shown great potential in medical image analysis. It allows to use multi-center training data with privacy protection. However, distributions local centers can vary from each other due different imaging vendors, and annotation protocols. Such variation degrades the performance of learning-based methods. To mitigate influence, two groups methods have been proposed for aims, i.e., global personalized The former are aimed improve a single model all test unseen (known as...