A5022722579- Retinal Imaging and Analysis
- Lipoproteins and Cardiovascular Health
- Retinal Diseases and Treatments
- Glaucoma and retinal disorders
- Sentiment Analysis and Opinion Mining
- Emotion and Mood Recognition
- Retinal and Optic Conditions
- Image Retrieval and Classification Techniques
- Health Systems, Economic Evaluations, Quality of Life
- Text and Document Classification Technologies
- Brain Tumor Detection and Classification
- Coronary Interventions and Diagnostics
- Metaheuristic Optimization Algorithms Research
- Natural Language Processing Techniques
- Monoclonal and Polyclonal Antibodies Research
- Face and Expression Recognition
- Digital Imaging for Blood Diseases
- Advanced Computing and Algorithms
- Visual Attention and Saliency Detection
- Statistical Methods in Clinical Trials
- Semantic Web and Ontologies
- Advanced Algorithms and Applications
- Corneal surgery and disorders
- Pharmaceutical Economics and Policy
- Ocular Diseases and Behçet’s Syndrome
Chinese Academy of Medical Sciences & Peking Union Medical College
2019-2024
Peking Union Medical College Hospital
2019-2024
Xidian University
2021-2024
Sichuan University
2024
Shenzhen University
2022-2023
University of Leicester
2023
George Mason University
2019
Amgen (United States)
2012-2018
Allergan (United States)
2018
Xi'an University of Technology
2018
<h3>Importance</h3> Reducing levels of low-density lipoprotein cholesterol (LDL-C) with intensive statin therapy reduces progression coronary atherosclerosis in proportion to achieved LDL-C levels. Proprotein convertase subtilisin kexin type 9 (PCSK9) inhibitors produce incremental lowering statin-treated patients; however, the effects these drugs on have not been evaluated. <h3>Objective</h3> To determine PCSK9 inhibition evolocumab patients. <h3>Design, Setting, and Participants</h3> The...
(Aim) Multiple sclerosis is a severe brain and/or spinal cord disease. It may lead to wide range of symptoms. Hence, the early diagnosis and treatment quite important. (Method) This study proposed 14-layer convolutional neural network, combined with three advanced techniques: batch normalization, dropout, stochastic pooling. The output pooling was obtained via sampling from multinomial distribution formed activations each region. In addition, we used data augmentation method enhance training...
Visual Emotion Analysis (VEA) aims at finding out how people feel emotionally towards different visual stimuli, which has attracted great attention recently with the prevalence of sharing images on social networks. Since human emotion involves a highly complex and abstract cognitive process, it is difficult to infer emotions directly from holistic or regional features in affective images. It been demonstrated psychology that are evoked by interactions between objects as well scenes within an...
Visual Emotion Analysis (VEA) aims at predicting people's emotional responses to visual stimuli. This is a promising, yet challenging, task in affective computing, which has drawn increasing attention recent years. Most of the existing work this area focuses on feature design, while little been paid dataset construction. In work, we introduce EmoSet, first large-scale emotion annotated with rich attributes, superior datasets four aspects: scale, annotation richness, diversity, and data...
A Large Language Model (LLM) tends to generate inconsistent and sometimes contradictory outputs when presented with a prompt that has equivalent semantics but is expressed differently from the original prompt. To achieve semantic consistency of an LLM, one key approaches finetune model prompt-output pairs semantically meanings. Despite its effectiveness, data-driven finetuning method incurs substantial computation costs in data preparation optimization. In this regime, LLM treated as ``black...
Large Language Models (LLMs) often generate inconsistent responses when prompted with semantically equivalent paraphrased inputs. Recently, activation steering, a technique that modulates LLM behavior by adjusting their latent representations during inference time, has been explored to improve the semantic consistency of LLMs. However, these methods typically operate at model component level, such as layer hidden states or attention heads. They face challenge due ``polysemanticity issue'',...
Interest in teleophthalmology has been growing, especially during the COVID-19 pandemic. The advent of fifth-generation (5G) wireless systems potential to revolutionize teleophthalmology, but these have not previously leveraged conduct therapeutic telemedicine ophthalmology field.To assess feasibility 5G real-time laser photocoagulation as a telemedicine-based treatment for diabetic retinopathy (DR).This was prospective study involving retinal specialist from Peking Union Medical College...
Visual Emotion Analysis (VEA) has attracted increasing attention recently with the prevalence of sharing images on social networks. Since human emotions are ambiguous and subjective, it is more reasonable to address VEA in a label distribution learning (LDL) paradigm rather than single-label classification task. Different from other LDL tasks, there exist intrinsic relationships between unique characteristics within them, as demonstrated psychological theories. Inspired by this, we propose...
To establish an automated method for identifying referable diabetic retinopathy (DR), defined as moderate nonproliferative DR and above, using deep learning-based lesion detection stage grading.A set of 12,252 eligible fundus images patients were manually annotated by 45 licenced ophthalmologists randomly split into training, validation, internal test sets (ratio 7:1:2). Another 565 consecutive clinical was established external set. For identification, four learning models programmed based...
Visual Emotion Analysis (VEA), which aims to predict people's emotions towards different visual stimuli, has become an attractive research topic recently. Rather than a single label classification task, it is more rational regard VEA as Label Distribution Learning (LDL) problem by voting from individuals. Existing methods often emotion distribution in unified network, neglecting the inherent subjectivity its crowd process. In psychology, Object-Appraisal-Emotion model demonstrated that each...
Purpose: To assess the accuracy and robustness of AI algorithm for detecting referable diabetic retinopathy (RDR), macular diseases (RMD), glaucoma suspect (GCS) from fundus images in community in-hospital screening scenarios. Methods We collected two color image datasets, namely, PUMCH (556 images, 166 subjects, four camera models) NSDE (534 134 models). The generates report after taking images. were labeled as RDR, RMD, GCS, or none three by 3 licensed ophthalmologists. resulting labels...
Purpose: To investigate choroidal changes in patients with Vogt-Koyanagi-Harada disease (VKH) using widefield swept-source optical coherence tomography angiography (SS-OCTA). Methods: In this cross-sectional study, 133 eyes of 69 VKH (52 28 active and 81 41 inactive patients) 104 52 age sex matched healthy volunteers were imaged a SS-OCTA instrument. On 12 mm × OCTA scans, mean thickness (MCT), vascularity index (CVI), choriocapillaris (CC) flow area, retinal (MRT) separately calculated the...
Visual emotion analysis (VEA) has attracted great attention recently, due to the increasing tendency of expressing and understanding emotions through images on social networks. Different from traditional vision tasks, VEA is inherently more challenging since it involves a much higher level complexity ambiguity in human cognitive process. Most existing methods adopt deep learning techniques extract general features whole image, disregarding specific evoked by various emotional stimuli....
Background: α4β7 integrin is a validated target in IBD. We studied abrilumab (AMG181/MEDI7183), fully human monoclonal antibody against integrin, subjects with ulcerative colitis (UC) and inadequate or loss of response to anti-tumor necrosis factors (TNFs) immunomodulators. Methods: This phase 2b, multicentre, randomised, double-blind, placebo (PBO)-controlled, parallel-group, multi-dose study enrolled moderate severe UC (total Mayo Score 6–12, rectosigmoidoscopy score ≥2). Subjects received...
To investigate whether stereoscopic vs. monoscopic viewing condition influences the evaluation of optic disc photographs for morphologic features and glaucoma likelihood in a general ophthalmologist population from multicenters on cloud-based platform.A cross-sectional study 519 pairs discs with adequate quality were selected presented using platform. A total 21 ophthalmologists 14 centers assessed 15 based 5R's rules estimated each assigned photograph. There 93 evaluated by panel...