A5100661279- Medical Image Segmentation Techniques
- Medical Imaging and Analysis
- 3D Shape Modeling and Analysis
- Advanced Neural Network Applications
- AI in cancer detection
- Advanced MRI Techniques and Applications
- Brain Tumor Detection and Classification
- Medical Imaging Techniques and Applications
- Cardiac Imaging and Diagnostics
- Image Retrieval and Classification Techniques
- Radiomics and Machine Learning in Medical Imaging
- Advanced X-ray and CT Imaging
- Infrared Target Detection Methodologies
- Image and Object Detection Techniques
- Linguistics and Cultural Studies
- Rough Sets and Fuzzy Logic
- Traditional Chinese Medicine Studies
- Retinal Imaging and Analysis
- Data Management and Algorithms
- COVID-19 diagnosis using AI
- Cancer-related molecular mechanisms research
- Cardiovascular Function and Risk Factors
- Computer Graphics and Visualization Techniques
- Advanced Measurement and Detection Methods
- Digital Image Processing Techniques
Beijing Normal University
2015-2024
Beijing Anzhen Hospital
2024
Capital Medical University
2024
Xi'an Technological University
2022
Hubei University of Technology
2018
Tongue diagnosis plays a pivotal role in traditional Chinese medicine (TCM) for thousands of years. As one the most important tongue characteristics, tooth-marked is related to spleen deficiency and can greatly contribute symptoms differentiation treatment selection. Yet, recognition TCM practitioners subjective challenging. Most previous studies have concentrated on subjectively selected features region gained accuracy under 80%. In present study, we proposed an artificial intelligence...
Whole heart segmentation is an important medical imaging method used to enable clinical applications. However, automatic of the still a challenging task due complexity and particularity images, especially when segmented into substructures. In this study, we present training strategy that relies on two-stage U-Net framework adaptive threshold window automatically segment whole its The consists region interest (ROI) detection accurate remove noisy parts data while preserving anatomical...
Deep learning (DL) has been widely used in biomedical image segmentation and automatic disease diagnosis, leading to state-of-the-art performance. However, automated cardiac diagnosis heavily relies on maps from magnetic resonance (CMR), most current DL methods, such as 2D convolution planes, 3D convolution, are not fully applicable CMR due loss of spatial structure information or large gap between slices. To make better exploit aspects the data improve accuracy, we propose a new structure,...
The recognition of tooth-marked tongues has important value for clinical diagnosis traditional Chinese medicine. Tooth-marked tongue is often related to spleen deficiency, cold dampness, sputum, effusion, and blood stasis. manifestations patients with include loss appetite, borborygmus, gastric distention, loose stool. Traditional conducted subjectively based on the doctor’s visual observation, its performance affected by subjectivity, experience, environmental lighting changes. In addition,...
The segmentation of coronary arteries is a vital process that helps cardiovascular radiologists detect and quantify stenosis. In this paper, we propose fully automated artery from cardiac data volume. method built on statistics region growing together with heuristic decision. First, the heart extracted using multi-atlas-based approach. Second, vessel structures are enhanced via 3D multiscale line filter. Next, seed points detected automatically through threshold preprocessing subsequent...
Cerebral vessel segmentation is essential and helpful for the clinical diagnosis related research. However, automatic of brain vessels remains challenging because variable shape high complex geometry. This study proposes a new active contour model (ACM) implemented by level-set method segmenting from TOF-MRA data. The energy function model, combining both region intensity boundary information, composed two terms, one term penalty term. global threshold representing lower gray target object...
Abstract Background Previous studies have shown the importance of energy deficiency and malfunctioning mitochondria in pathophysiology hypertrophic cardiomyopathy (HCM). There has been a little research into relationship between plasma free fatty acids (FFA), one heart’s main sources, HCM. We evaluated its clinical HCM to see if there was link FFA metabolism Methods In single-center retrospective observational study, we investigated 420 patients diagnosed at Beijing Anzhen Hospital January...
Abstract Background Cervical cancer cell detection is an essential means of cervical screening. However, for thin-prep cytology test (TCT)-based images, the accuracies traditional computer-aided algorithms are typically low due to overlapping cells with blurred cytoplasmic boundaries. Some typical deep learning-based methods, e.g., ResNets and Inception-V3, not always efficient images differences between natural images. As a result, these networks difficult directly apply clinical practice...
Brain vessel segmentation is a fundamental component of cerebral disease screening systems. However, detecting vessels still challenging task owing to their complex appearance and thinning geometry as well the contrast decrease from root its thin branches. We present method for vasculature in Magnetic Resonance Angiography (MRA) images. First, we apply volume projection, 2D segmentation, back-projection procedures first stage background subtraction reservation. Those labeled or voxels are...
Segmentation is one of the most challenging problems in field medical image analysis, and blood vessels are especially difficult to extract. In this paper, we propose a novel method for segmentation cerebral from magnetic resonance angiography (MRA) images based on Frangi's vesselness measure ball B-Spline. First, apply find putative centerlines tubular structures along with their estimated radii. Then B-Spline adopted construct 3D vascular trees. Results head MRA datasets demonstrate...
Abstract The pedestrian re-identification task explored in this study has strong application scenarios real life. Generally, actual scenarios, the samples taken by camera are unlabeled, so pseudo-labeling is used. However, low-quality image feature extraction and imperfect clustering will produce incorrect results, which have a negative impact on performance. This article proposes dual method (PLR). core idea of to explore distribution differences similarities between two data sets through...
Abstract In the process of UAV target recognition, is small, feature not obvious, and recognition accuracy low, a detection method called YOLOv8-UD proposed based on YOLOv8. This replaces downsampling layers with SPD convolutions to mitigate fine-grained loss issues. It employs BiFPN structure get information which multi-scale, enhancing representation global semantic message Improve awareness small goals. As result, detection, it significantly enhances correctness effectiveness....