A5070980467- Radiomics and Machine Learning in Medical Imaging
- Atmospheric aerosols and clouds
- AI in cancer detection
- Thyroid Cancer Diagnosis and Treatment
- Solar Radiation and Photovoltaics
- Atmospheric chemistry and aerosols
- Meteorological Phenomena and Simulations
- Parallel Computing and Optimization Techniques
- Advanced Neural Network Applications
- Atmospheric Ozone and Climate
- Thyroid and Parathyroid Surgery
- Advanced Data Storage Technologies
- Music and Audio Processing
- Medical Image Segmentation Techniques
- Domain Adaptation and Few-Shot Learning
- Anomaly Detection Techniques and Applications
- Brain Tumor Detection and Classification
- Fetal and Pediatric Neurological Disorders
- Advanced Image Fusion Techniques
- Single-cell and spatial transcriptomics
- Advanced Sensor and Control Systems
- Embedded Systems Design Techniques
- Combustion and flame dynamics
- Generative Adversarial Networks and Image Synthesis
- Cell Image Analysis Techniques
Sichuan University
2018-2025
Guangdong Urban & Rural Planning and Design Institute
2025
National Supercomputing Center in Wuxi
2025
Jilin Jianzhu University
2021-2024
Liaocheng People's Hospital
2017-2024
Fuzhou University
2023-2024
New York University
2023-2024
Shandong First Medical University
2020-2024
Intel (United States)
2024
Nanjing University of Information Science and Technology
2019-2023
Medical image segmentation is a critical component in clinical practice, facilitating accurate diagnosis, treatment planning, and disease monitoring. However, current methods predominantly rely on customized models, which exhibit limited generality across diverse tasks. In this study, we present MedSAM, the inaugural foundation model designed for universal medical segmentation. Harnessing power of meticulously curated dataset comprising over one million images, MedSAM not only outperforms...
A deep-learning-based cloud detection and classification algorithm for advanced Himawari imager (AHI) measurements from the geostationary satellite Himawari-8 has been developed. It is found that a combination of observed radiances simulated clear-sky can substantially improve phase discrimination, especially optically thin clouds. Therefore, detection, classification, multilayer are obtained simultaneously multispectral using deep neural networks (DNNs). Two DNN models established all-day...
As people's awareness of ecological protection increases, bird sound monitoring has received more and attention. Among them, using as part audio recognition become a hot research topic. Since sounds are usually collected in natural environments, they contain lot noise, which will affect the results. To solve this problem, paper designs Convolutional Recurrent Network (CRN) that enhances feature representation along frequency axis. This method is based on Short-time Fourier transform (STFT)...
Abstract A particle swarm optimization-back propagation neural network (PSO-BP) is proposed. First, we collect and preprocess the planning data for low-carbon landscape environment to ensure accuracy consistency of data; then propose PSO-BP model that combines global optimization characteristics algorithm nonlinear mapping capability backpropagation network. Empirical studies show this can effectively reduce energy consumption carbon emissions, thus improving ecological quality, improve...
Abstract In this study, we propose a novel joint training model for named entity recognition (NER) that combines BERT, BiLSTM, CRF, and reading comprehension (RC) mechanism. Traditional BERT‐BiLSTM‐CRF models often struggle with inaccurate boundary detection excessive fragmentation of entities due to their lack specialized vocabulary. Our addresses these issues by integrating an RC mechanism, which helps refine fragmented results enabling the more precisely identify boundaries without...
BACKGROUND: Automatic segmentation of individual tooth root is a key technology for the reconstruction three-dimensional dental model from Cone Beam Computed Tomography (CBCT) images, which great significance orthodontic, implant and other diagnosis treatment planning. OBJECTIVES: Currently, mainly done manually because similar gray alveolar bone CBCT images. This study aims to explore automatic algorithm axial image sequence based on deep learning. METHODS: We proposed new method learning...
Cell segmentation is a critical step for quantitative single-cell analysis in microscopy images. Existing cell methods are often tailored to specific modalities or require manual interventions specify hyper-parameters different experimental settings. Here, we present multi-modality benchmark, comprising over 1500 labeled images derived from more than 50 diverse biological experiments. The top participants developed Transformer-based deep-learning algorithm that not only exceeds existing but...
Abstract Background Ultrasound three-dimensional visualization, a cutting-edge technology in medical imaging, enhances diagnostic accuracy by providing more comprehensive and readable portrayal of anatomical structures compared to traditional two-dimensional ultrasound. Crucial this visualization is the segmentation multiple targets. However, challenges like noise interference, inaccurate boundaries, difficulties segmenting small exist multi-target ultrasound images. This study, using neck...
Abstract. Four distinct retrieval algorithms, comprising two physics-based and machine-learning (ML) approaches, have been developed to retrieve cloud base height (CBH) its diurnal cycle from Himawari-8 geostationary satellite observations. Validations conducted using the joint CloudSat/CALIOP (Cloud-Aerosol Lidar with Orthogonal Polarization) CBH products in 2017, ensuring independent assessments. Results show that ML-based algorithms exhibit markedly superior performance (with a...
Clouds play an important role in the Earth's climate system; however, various observational methods describe clouds differently, leading to cloud products being described with different characteristics, and affecting our understanding of effects. To address this problem, study integrates into transfer-learning procedure a deep learning model determined Cloud Effective Radius (CER), Optical Thickness (COT), Top Height (CTH) from Himawari-8 thermal infrared measurements. The retrieval results...
Deep-learning based generative models are proven to be capable for achieving excellent results in numerous image processing tasks with a wide range of applications. One significant improvement deep-learning approaches compared traditional is their ability regenerate semantically coherent images by only relying on an input limited information. This advantage becomes even more crucial when the size very minor proportion output size. Such expansion can challenging as missing area may originally...
Abstract Purposes The incidence of thyroid cancer has increased annually, and a heavy psychological economic burden on society individuals. Based data from patients treated in Liaocheng People's Hospital 2015 to 2018, with Chinese national regional characteristics, this study, we addressed the controversy which initial surgical mode, lobectomy or total thyroidectomy, is most effective. Methods Clinical pathological 2108 cancer, who were initially diagnosed surgically, collected Department...
Breast cancer has the highest prevalence rate among women worldwide. Early detection of breast is crucial for successful treatment and reducing mortality rate. However, tumor ultrasound (US) image still a challenging work in computer-aided diagnosis (CAD).This study aims to develop novel automated algorithm based on deep learning.We proposed new learning network named One-step model which have one input two outputs, first was segmentation result other used false-positive reduction. The...
BACKGROUND: Thyroid ultrasonography is widely used to diagnose thyroid nodules in clinics. Automatic localization of can promote the development intelligent diagnosis and reduce workload radiologists. However, besides ultrasound image has low contrast high noise, are diverse shape vary greatly size. Thus, nodule detection images still a challenging task. OBJECTIVE: This study proposes an automatic algorithm locate B Doppler images. method be screen provide basis for subsequent segmentation...