A5112684346- Gene expression and cancer classification
- Bioinformatics and Genomic Networks
- Molecular Biology Techniques and Applications
- Gene Regulatory Network Analysis
- Metabolomics and Mass Spectrometry Studies
- Spectroscopy and Chemometric Analyses
- Radiomics and Machine Learning in Medical Imaging
- Machine Learning in Bioinformatics
- Single-cell and spatial transcriptomics
- Blind Source Separation Techniques
- AI in cancer detection
- Cell Image Analysis Techniques
- Analytical Chemistry and Chromatography
- Advanced Chemical Sensor Technologies
- Medical Image Segmentation Techniques
- Cancer Genomics and Diagnostics
- Speech and Audio Processing
- Image Processing Techniques and Applications
- MRI in cancer diagnosis
- Medical Imaging Techniques and Applications
- Genomics and Chromatin Dynamics
- Sparse and Compressive Sensing Techniques
- Evolutionary Algorithms and Applications
- RNA and protein synthesis mechanisms
- Face and Expression Recognition
Shandong University of Technology
2025
Virginia Tech
2015-2024
Beijing Institute of Technology
2024
Alibaba Group (China)
2024
Winston-Salem/Forsyth County Schools
2023
University of Minnesota System
2021
Cedars-Sinai Medical Center
2021
Wake Forest University
2021
Beijing Information Science & Technology University
2018
Dartmouth College
2016
The inability to detect premature atherosclerosis significantly hinders implementation of personalized therapy prevent coronary heart disease. A comprehensive understanding arterial protein networks and how they change in early could identify new biomarkers for disease detection improved therapeutic targets.
Identifying urban functional zones is an important task in planning and smart city construction. Accurately identifying analyzing their spatial distribution crucial for optimizing layout, improving management balancing human–environment interaction. However, most of the existing studies focus on analysis individual data sources, which have difficulty fully reflecting complex structures differences cities. To solve this problem, paper proposes a new method area identification integrates...
Echo state networks (ESNs) are a novel form of recurrent neural (RNNs) that provide an efficient and powerful computational model approximating nonlinear dynamical systems. A unique feature ESN is large number neurons (the "reservoir") used, whose synaptic connections generated randomly, with only the from reservoir to output modified by learning. Why randomly fixed RNN gives such excellent performance in systems still not well understood. In this brief, we apply random matrix theory examine...
Currently, cancer therapy remains limited by a “one-size-fits-all” approach, whereby treatment decisions are based mainly on the clinical stage of disease, yet fail to reference individual's underlying biology and its role driving malignancy. Identifying better personalized therapies for is hindered lack high-quality “omics” data sufficient size produce meaningful results ability integrate biomedical from disparate technologies. Resolving these issues will help translation research clinic...
Abstract Summary: We develop a novel unsupervised deconvolution method, within well-grounded mathematical framework, to dissect mixed gene expressions in heterogeneous tumor samples. implement an R package, UNsupervised DecOnvolution (UNDO), that can be used automatically detect cell-specific marker genes (MGs) located on the scatter radii of expressions, estimate cellular proportions each sample and deconvolute into expression profiles. demonstrate performance UNDO over wide range...
Lack of understanding endocrine resistance remains one the major challenges for breast cancer researchers, clinicians, and patients. Current reductionist approaches to molecular signaling driving have offered mostly incremental progress over past 10 years. As field systems biology has begun mature, network modeling tools being developed applied therein offer a different way think about how regulation critical cellular functions are integrated. To gain novel insights, we first describe some...
Three-dimensional (3D) volumetric neural image segmentation is crucial to reconstructing accurate neuron structures. However, due the structural complexity of neurons and diverse imaging qualities microscopes, it challenging achieve both accuracy efficiency. In this paper, we propose a teacher-student learning framework for fast segmentation. The inference performed using light-weighted student network which benefits from knowledge distillation teacher with higher capacity. Evaluated on...
Abstract Missing values are a major issue in quantitative proteomics analysis. While many methods have been developed for imputing missing high-throughput data, comparative assessment of imputation accuracy remains inconclusive, mainly because mechanisms contributing to true complex and existing evaluation methodologies imperfect. Moreover, few studies provided an outlook future methodological development. We first re-evaluate the performance eight representative targeting three typical...
The key to geospatial data integration lies in identifying corresponding objects from different sources. Aiming at the problem of low matching accuracy entities under a single feature attribute, entity method based on multi-feature value calculation is proposed. Firstly, when dealing with POI (point interest) data, similarity terms name, address, and distance calculated by combining improved hybrid method, Jaccard Euclidean metric method. Secondly, random forest algorithm utilized...
Established human breast cancer cell lines are widely used as experimental models in research. While these and their variants share many phenotypic characteristics with tumors, the extent to which they reflect underlying molecular biology of remains controversial. We explored this issue using a probabilistic rather than heuristic approach. Data from gene expression microarrays were compare global structures transcriptomes three estrogen receptor alpha positive (ER+) (MCF-7, T47D, ZR-75-1) 13...
Identification of differentially expressed subnetworks from protein–protein interaction (PPI) networks has become increasingly important to our global understanding the molecular mechanisms that drive cancer. Several methods have been proposed for PPI subnetwork identification, but dependency among network member genes is not explicitly considered, leaving many hub largely unidentified. We present a new method, based on bagging Markov random field (BMRF) framework, improve identification...
Segmenting cell nuclei in microscopic images has become one of the most important routines modern biological applications. With vast amount data, automatic localization, i.e. detection and segmentation, is highly desirable compared to time-consuming manual processes. However, automated segmentation challenging due large intensity inhomogeneities background. We present a new method for progressive localization using data-adaptive models that can better handle inhomogeneity problem. perform...
While non-negative blind source separation (nBSS) has found many successful applications in science and engineering, model order selection, determining the number of sources, remains a critical yet unresolved problem. Various selection methods have been proposed applied to real-world data sets but with limited success, both over- under-estimation reported. By studying existing schemes, we that unsatisfactory results are mainly due invalid assumptions, oversimplification, subjective...
Designing a single automatic and accurate segmentation approach for different classes of white blood cells is challenging task. This paper presents fully automated framework to segment both nuclei cytoplasm five major in the peripheral smears based on color texture enhancement. Particularly, new gray-scale transform generated three representative channels separate from background by Poisson distribution minimum error thresholding. For segmentation, discrete wavelet (DWT) morphological...
Bioimage classification is a fundamental problem for many important biological studies that require accurate cell phenotype recognition, subcellular localization, and histopathological classification. In this paper, we present new bioimage method can be generally applicable to wide variety of problems. We propose use high-dimensional multi-modal descriptor combines multiple texture features. also design novel subcategory discriminant transform (SDT) algorithm further enhance the...
Among multiple subtypes of tissue or cell, subtype-specific differentially-expressed genes (SDEGs) are defined as being most-upregulated in only one subtype but not any other. Detecting SDEGs plays a critical role the molecular characterization and deconvolution multicellular complex tissues. Classic differential analysis assumes null hypothesis whose test statistic is subtype-specific, thus can produce high false positive rate and/or lower detection power. Here we first introduce...
Ideally, a molecularly distinct subtype would be composed of molecular features that are expressed uniquely in the interest but no others-so-called marker genes (MGs). MG plays critical role characterization, classification or deconvolution tissue cell subtypes. We and others have recognized test statistics used by most methods do not exactly satisfy definition often identify inaccurate MG.We report an efficient accurate data-driven method, formulated as Cosine-based One-sample Test (COT)...