A5077384503- Medical Image Segmentation Techniques
- AI in cancer detection
- Advanced Neural Network Applications
- Radiomics and Machine Learning in Medical Imaging
- Medical Imaging Techniques and Applications
- 3D Shape Modeling and Analysis
- Medical Imaging and Analysis
- Advanced MRI Techniques and Applications
- COVID-19 diagnosis using AI
- Advanced X-ray and CT Imaging
- Image Retrieval and Classification Techniques
- Generative Adversarial Networks and Image Synthesis
- Diabetic Foot Ulcer Assessment and Management
- Lung Cancer Diagnosis and Treatment
- Computer Graphics and Visualization Techniques
- Advanced Neuroimaging Techniques and Applications
- Functional Brain Connectivity Studies
- Digital Imaging for Blood Diseases
- Cardiac Imaging and Diagnostics
- Dental Radiography and Imaging
- Domain Adaptation and Few-Shot Learning
- Human Pose and Action Recognition
- Brain Tumor Detection and Classification
- Infrared Thermography in Medicine
- Advanced Numerical Analysis Techniques
Princeton University
2019-2024
Rutgers Sexual and Reproductive Health and Rights
2021
National Heart Centre Singapore
2021
Shanghai Artificial Intelligence Laboratory
2021
Adobe Systems (United States)
2021
Hong Kong Science and Technology Parks Corporation
2020
Rutgers, The State University of New Jersey
2011-2019
Group Sense (China)
2019
Siemens Healthcare (United States)
2018
University of North Carolina at Chapel Hill
2017
In general image recognition problems, discriminative information often lies in local patches. For example, most human identity exists the patches containing faces. The same situation stays medical images as well. "Bodypart identity" of a transversal slice-which bodypart slice comes from-is indicated by information, e.g., cardiac and an aorta arch are only differentiated mediastinum region. this work, we design multi-stage deep learning framework for classification apply it on recognition....
Nuclei segmentation is a fundamental task in histopathology image analysis. Typically, such tasks require significant effort to manually generate accurate pixel-wise annotations for fully supervised training. To alleviate tedious and manual effort, this paper we propose novel weakly framework based on partial points annotation, i.e., only small portion of nuclei locations each are labeled. The consists two learning stages. In the first stage, design semi-supervised strategy learn detection...
Abstract Breast carcinoma is the most common cancer among women worldwide that consists of a heterogeneous group subtype diseases. The whole-slide images (WSIs) can capture cell-level heterogeneity, and are routinely used for diagnosis by pathologists. However, key driver genetic mutations related to targeted therapies identified genomic analysis like high-throughput molecular profiling. In this study, we develop deep-learning model predict biological pathway activities directly from WSIs....
Transformers have demonstrated remarkable performance in natural language processing and computer vision. However, existing vision struggle to learn from limited medical data are unable generalize on diverse image tasks. To tackle these challenges, we present MedFormer, a data-scalable Transformer designed for generalizable 3D segmentation. Our approach incorporates three key elements: desirable inductive bias, hierarchical modeling with linear-complexity attention, multi-scale feature...
With the rise of social media and Location-Based Social Networks (LBSN), check-in data across platforms has become crucial for User Identity Linkage (UIL). These not only reveal users' spatio-temporal information but also provide insights into their behavior patterns interests. However, cross-platform identity linkage faces challenges like poor quality, high sparsity, noise interference, which hinder existing methods from extracting user information. To address these issues, we propose a...
Accurate localization of the anatomical landmarks on distal femur bone in 3D medical images is very important for knee surgery planning and biomechanics analysis. However, landmark identification process often conducted manually or by using inserted auxiliaries, which time-consuming lacks accuracy. In this paper, an automatic method proposed to determine positions initial geometric surface MR images. Based results from convolutional neural network (CNN) classifiers shape statistics, we use...
Automatic and accurate 3D cardiac image segmentation plays a crucial role in disease diagnosis treatment. Even though CNN based techniques have achieved great success medical segmentation, the expensive annotation, large memory consumption, insufficient generalization ability still pose challenges to their application clinical practice, especially case of from high-resolution large-dimension volumetric imaging. In this paper, we propose few-shot learning framework by combining ideas...
Overcoming barriers on the use of multi-center data for medical analytics is challenging due to privacy protection and heterogeneity in healthcare system. In this study, we propose Distributed Synthetic Learning (DSL) architecture learn across multiple centers ensure sensitive personal information. DSL enables building a homogeneous dataset with entirely synthetic images via form GAN-based learning. The proposed has following key functionalities: multi-modality learning, missing modality...
Cardiac tagging magnetic resonance imaging (t-MRI) is the gold standard for regional myocardium deformation and cardiac strain estimation. However, this technique has not been widely used in clinical diagnosis, as a result of difficulty motion tracking encountered with t-MRI images. In paper, we propose novel deep learning-based fully unsupervised method vivo on We first estimate field (INF) between any two consecutive frames by bi-directional generative diffeomorphic registration neural...
Limb salvage surgery of malignant pelvic tumors is the most challenging procedure in musculoskeletal oncology due to complex anatomy bones and soft tissues. It crucial accurately resect with appropriate margins this procedure. However, there still a lack efficient repetitive image planning methods for tumor identification segmentation many hospitals. In paper, we present novel deep learning-based method segment bone MRI. Our uses multi-view fusion network extract pseudo-3D information from...
Fully convolutional network (FCN) has shown potency in segmenting heterogeneous objects from natural images with high run-time efficiency. This technique, however, is not able to produce continuous, smooth and shape-preserved regions consistently due complex organ structures occasional weak appearance information commonly observed various anatomical medical images. In this paper, we propose a deep end-to-end two task-specific branches ensure continuousness, smoothness shape-preservation...
Detecting deception in interpersonal dialog is challenging since deceivers take advantage of the give-and-take interaction to adapt any sign skepticism an interlocutor's verbal and nonverbal feedback. Human detection accuracy poor, often with no better than chance performance. In this investigation, we consider whether automated methods can produce results if emphasizing possible disruption interactional synchrony signal interactant truthful or deceptive. We propose a data-driven unobtrusive...
Accurate segmentation of the 30+ subcortical structures in MR images whole diseased brains is challenging due to inter-subject variability and complex geometry brain anatomy. However a clinically viable solution yielding precise would enable: 1) accurate, objective measurement structure volumes many which are associated with diseases such as Alzheimer's, 2) therapy monitoring 3) drug development. Our contributions two-fold. First we construct an extended adaptive statistical atlas method...