A5103198665- Image Retrieval and Classification Techniques
- Gaussian Processes and Bayesian Inference
- Image and Object Detection Techniques
- AI in cancer detection
- Advanced Image and Video Retrieval Techniques
- Statistical Methods and Inference
- Model Reduction and Neural Networks
- Medical Image Segmentation Techniques
- Brain Tumor Detection and Classification
- Markov Chains and Monte Carlo Methods
- Retinal Imaging and Analysis
- Machine Learning and Algorithms
- Advanced Measurement and Detection Methods
- Advanced Algorithms and Applications
- Artificial Intelligence in Healthcare
- Time Series Analysis and Forecasting
- Machine Fault Diagnosis Techniques
- Image and Signal Denoising Methods
- Bayesian Methods and Mixture Models
- Fault Detection and Control Systems
- Topic Modeling
- Remote Sensing in Agriculture
- Air Quality Monitoring and Forecasting
- Smart Agriculture and AI
- Seismic Imaging and Inversion Techniques
Shanghai Jiao Tong University
2009-2024
Tianjin Medical University Cancer Institute and Hospital
2023-2024
Purdue University West Lafayette
2018-2024
Nanjing University of Information Science and Technology
2022-2024
University of Pennsylvania
2024
Hainan University
2024
Suzhou Polytechnic Institute of Agriculture
2022-2024
Georgia Institute of Technology
2022-2023
PLA Air Force Aviation University
2013-2023
Chinese Academy of Forestry
2023
Abstract Cancer of unknown primary (CUP) site poses diagnostic challenges due to its elusive nature. Many cases CUP manifest as pleural and peritoneal serous effusions. Leveraging cytological images from 57,220 at four tertiary hospitals, we developed a deep-learning method for tumor origin differentiation using histology (TORCH) that can identify malignancy predict in both hydrothorax ascites. We examined performance on three internal ( n = 12,799) two external 14,538) testing sets. In...
In recent years, extreme natural hazards threaten cities more than ever due to contemporary society’s high vulnerability in cities. Hence, local governments need implement risk mitigation and disaster operation management enhance resilience Transforming existing open spaces within into emergency shelters is an effective method of providing essential life support agent recovery the wake disasters. Emergency planning must identify suitable locations for reasonably allocate evacuees those...
Abstract Background Karst vegetation is of great significance for ecological restoration in karst areas. Vegetation Indices (VIs) are mainly related to plant yield which helpful understand the status Recently, surveys have gradually shifted from field remote sensing-based methods. Coupled with machine learning methods, Unmanned Aerial Vehicle (UAV) multispectral sensing data can effectively improve detection accuracy and extract important spectrum features. Results In this study, UAV image...
The online detection of a very long and weak chirp signal is studied. has an extremely slowly decreasing frequency, corrupted by white Gaussian noise possibly also powerful tones. By exploring comparing candidate methods, it found that the Hough transform (HT) detector appears to be most suitable given constraints on computational load detectability. analytical simulational performance HT are obtained compared with generalized likelihood ratio test (GLRT), which assumed optimal. Applying...
Summary Stochastic gradient Markov chain Monte Carlo algorithms have received much attention in Bayesian computing for big data problems, but they are only applicable to a small class of problems which the parameter space has fixed dimension and log-posterior density is differentiable with respect parameters. This paper proposes an extended stochastic algorithm which, by introducing appropriate latent variables, can be applied more general large-scale such as those involving jumping missing...
Deep learning has been the engine powering many successes of data science. However, deep neural network (DNN), as basic model learning, is often excessively over-parameterized, causing difficulties in training, prediction and interpretation. We propose a frequentist-like method for sparse DNNs justify its consistency under Bayesian framework: proposed could learn DNN with at most O(n/log(n)) connections nice theoretical guarantees such posterior consistency, variable selection asymptotically...
Abstract While fiducial inference was widely considered a big blunder by R.A. Fisher, the goal he initially set—‘inferring uncertainty of model parameters on basis observations’—has been continually pursued many statisticians. To this end, we develop new statistical method called extended Fiducial (EFI). The achieves leveraging advanced computing techniques while remaining scalable for data. Extended involves jointly imputing random errors realized in observations using stochastic gradient...
In this paper, a novel liver lesion diagnosis approach based on multi-phase enhanced CT images is proposed. Regions of Interest (ROIs) which are drawn by an experienced radiologist categorized into 4 classes: normal, cyst, haemangioma and hepatic cellular carcinoma. The scheme includes 3 steps: feature extraction, selection classification. For each ROI, distinct kinds features extracted using First Order Statistics (FOS), Second (SGLCM), Temporal Features, where 5 different sets constructed...
While recent research in music generation has mostly focused on encoder decoder architectures and self-attention mechanisms, prominent advancements regarding GANs have not yet been incorporated for the creation of music. These include solutions major challenges when training GANs, most importantly instability. In this work, we aim to apply new knowledge generation, order make it more efficient enable automatic higher quality. We utilize progressive approach towards implement train symbolic...
Computational pathology for gigapixel whole-slide images (WSIs) at slide level is helpful in disease diagnosis and remains challenging. We propose a context-aware approach termed WSI inspection via transformer (WIT) slide-level classification holistically modeling dependencies among patches on WSI. WIT automatically learns feature representation of by aggregating features all image patches. evaluate performance state-of-the-art baseline method. achieved an accuracy 82.1% (95% CI,...
In multivariate and multistep time series prediction research, we often face the problems of insufficient spatial feature extraction time-dependent mining historical data, which also brings great challenges to analysis prediction. Inspired by attention mechanism residual module, this study proposes a method based on convolutional-residual gated recurrent hybrid model (CNN-DA-RGRU) with two-layer solve problem in these two stages. Specifically, convolution module proposed is used extract...
OCT images have now become a very popular topic in the field of image processing. By measuring retinal nerve fiber layer thickness, such diseases like glaucoma or cataract can be diagnosed. An automated boundary segmentation algorithm is proposed for fast and reliable quantification six intra-retinal boundaries optical coherence tomography (OCT) images. The includes four steps. First all, will filtered by bilateral filter which suppress local noise but keep global variation across boundary....
Exactly extracting the stable feature of high resolution SAR image as well matching it are two critical steps for Antomatic regiestation systems. It is suggested that Scale Invariant Feature Transform (SIFT) algorithm can be applied in optical registration systems and four representative experiments were performed to test its validity. found SIFT accurately register images than traditional Harris applicability precision.
Near-real time estimation of precipitation from geostationary satellites plays a vital role in natural disaster mitigation due to timely monitoring, high spatial-temporal resolution and large coverage, yet this research remains challenge. In research, novel Deep Learning-based algorithm entitled Precipitation Estimation using Multi-Scale network (DLPE-MS) is proposed estimate during summer over eastern Continental United States (CONUS) America. When inputting bispectral satellite information...
In the post-epidemic era, online teaching has become normal mode. The blended mode which combines and offline efforts integrates complements main direction of education development in colleges universities. new model puts forward higher requirements for quality assurance to improve Based on deep learning-oriented evaluation index weight coefficient algorithm, this paper analyzes learning data supervision feedback suggestions smart platform spring 2020 2022 our school, constructs an system...
The online detection of a very long and weak chirp signal is, studied. has an extremely slowly-decreasing frequency is corrupted by white Gaussian noise, also possibly powerful tones. Four methods (the Hough transform, multiple tracker, Page's test EM algorithm) are explored. It found that the transform (HT) detector appears to be most suitable given constraints on computational load detectability. compared with GLRT, which assumed as "optimal" possible. Applying threshold for HT can...
Stochastic gradient Markov chain Monte Carlo (MCMC) algorithms have received much attention in Bayesian computing for big data problems, but they are only applicable to a small class of problems which the parameter space has fixed dimension and log-posterior density is differentiable with respect parameters. This paper proposes an extended stochastic MCMC lgoriathm which, by introducing appropriate latent variables, can be applied more general large-scale such as those involving jumping...
The aim of the present study is to build a software implementation previous and diagnose discoid lateral menisci on knee joint radiograph images. A total 160 images from normal individuals patients who were diagnosed with included. Our includes two parts: preprocessing measurement. In first phase, whole image was analyzed obtain basic information about patient. Machine learning used segment original image. Image enhancement denoising tools strengthen remove noise. second edge detection...