Research Projects

The central theme of the lab is focused on developing computational and informatics methods for integrative analysis of multimodal imaging data, high throughput “omics” data, cognitive and other biomarker data, electronic health record data, and rich biological knowledge such as pathways and networks, with applications to various complex disorders. Our research interests include medical image computing, bioinformatics, machine learning, network science, visual analytics, and big data science in biomedicine. The following are some of our research activities.

Deep Learning for Multidimensional Brain Imaging Genetics

Developing novel deep learning and bioinformatics algorithms and tools for comprehensive joint analysis of large-scale heterogeneous imaging genomics data, using Alzheimer’s Disease Neuroimaging Initiative (ADNI) database as a test bed. Integrating deep learning models with canonical correlation association for identifying correlated imaging and genetic features in the multi-view manner.

Integrative Bioinformatics Approaches to Human Brain Genomics and Connectomics

Aims: (1) develop novel computational pipeline for systematic characterization of structural connectome optimized for imaging genomics; (2) develop novel bioinformatics strategies to determining genetic basis of structural connectome. Using the Alzheimer’s Disease Neuroimaging Initiative (ADNI) and Human Connectome Project (HCP) cohorts as test beds to develop methods and tools with potential for a better understanding of the interplay between genes, brain connectivity and function.

Network-based GWAS for Identifying Tissue-specific Functional Module

Investigate the role of genetic variation in disordered brain structure and function using neuroimaging and biomarkers as phenotypes. Proposed two new module identification frameworks: (i) a machine learning based approach and (ii) a GWAS top-neighbor-based (tnGWAS) module identification approach, to mine the phenotype-specific gene set (GS)-imaging Quantitative Trait (iQT) associations by integrating tissue-specific functional interaction network with GWAS results of corresponding phenotype.

New paradigm of Imaging Genetic Enrichment Analysis (IGEA)

Explore the interlinked genetic markers and correlated brain phenotypes from large scale imaging genetic data. Proposed a two-dimensional enrichment paradigm-Imaging Genetic Enrichment Analysis (IGEA)-to jointly consider meaningful gene sets (GS) and brain circuits (BC) and examine whether any given GS–BC pair is enriched in a list of gene-iQT findings.

Graphic Mining of High-order Drug interactions and Directional Effects on Myopathy

Using electronic medical records, traditional Adverse Drug Effects (ADEs) analysis focus on either mining single drug-single ADE association, or multi-drug interactions but ignore direction of drug interaction. Proposed to examine directional high-order Drug-Drug Interactions (DDI) on ADEs using Electronic Medical Records based on the hypothesis that ADE risk of drug1 versus (drug1, drug2) is different from ADE risk of drug2 versus (drug1, drug2).

Scientific Mapping and Visualization of ADNI

Alzheimer's Disease Neuroimaging Initiative (ADNI) was launched to test if MRI, PET, other biological markers, and clinical and neuropsychological assessment can be combined to measure the progression of mild cognitive impairment (MCI) and early Alzheimer's disease (AD). This project investigated papers using ADNI data, to visualize scientific growth and impact of the ADNI efforts using various types of information including co-interaction network, co-authorship network, geospatial distribution, publication distribution, change of research field, et al.

Multivariate GWAS on Amyloid Imaging Phenotypes

Jointly analyzed multiple related traits and aimed to discover not only quantitative trait loci (QTLs) specific to individual traits, but also QTLs implicated jointly by multiple traits. Performed multivariate GWAS on amyloid imaging phenotypes and compared its performance with that of univariate GWAS.

Image computing methods in biomedicine

With the continuous development and progress of medical imaging technology and computer technology, medical image analysis has become an indispensable tool and technical means in medical research, clinical disease diagnosis and treatment. Therefore, a large number of image data such as MRI, pet, CT and other three-dimensional images can be used for research. This work aims to study and apply new graphic and image calculation and analysis methods, and design a series of tools and algorithms for detecting brain loss related to hippocampal atrophy, cortical analysis of autism, multiple sclerosis related to thalamic atrophy, craniofacial malformation and cardiac motor analysis.

Statistical analysis method based on image features

Medical image analysis has been widely used in clinical auxiliary screening, diagnosis, grading, treatment decision-making and guidance, curative effect evaluation and so on. The purpose of this work is to carry out the early diagnosis of Alzheimer's disease through the surface morphological analysis of three-dimensional images, and to study new statistical methods to enhance the statistical ability.

Multimodal data mining and biometric extraction

Medical image classification and recognition, location and detection, tissue and organ and lesion segmentation are the main application fields of deep learning methods in medical image analysis. There are great differences between medical image analysis with different imaging principles and natural image analysis in the field of computer vision, so we can solve a problem from different angles and dimensions. The purpose of this work is to identify biometrics from multidimensional data sets (including multimodal imaging data, high-throughput omics data and fluid biomarker data) to predict cognitive level and diagnose cognitive defects.

Genetic and multiomic analysis of phenotypes in Alzheimer's disease

The purpose of this study was to use neuroimaging and biomarker features as phenotypes to detect the role of genetic variation in brain dysfunction. We also designed advanced computational methods for genetic analysis of quantitative phenotypes in Alzheimer's disease. In view of the broad prospects of ADNI in the field of multiomics (for example, including data from genome, epigenome, transcriptome, proteome and metabolome), we are interested in expanding the scope of imaging omics research from the field of genome to the field of multiomics.