Vince D. Calhoun, PhD
 |
Professor, Wallace H. Coulter Department of Biomedical Engineering, Emory University & Georgia Institute of Technology
Distinguished University Professor, Psychology, Georgia State University
Founding Director, Center for Translational Research in Neuroimaging and Data Science (TReNDS)
Professor, Depts. of Neurology, Psychiatry, and Pediatrics
Professor, School of Electrical and Computer Engineering, College of Engineering, Georgia Institute of Technology
Graduate Programs
- Full Member - Neuroscience
Education
PhD, University of Maryland, 2002MS, Johns Hopkins University, 1996
MA, Johns Hopkins University, 1993
BS, University of Kansas, 1991
Contact Information
Email: vince.calhoun@emory.edu
Address:
Center for Translational Research in Neuroimaging and Data Science
55 Park Place NE, 18th Floor
Atlanta, GA 30303
In my lab we have been focused on ways to apply engineering principles to extract knowledge from brain imaging data through developing and optimizing methods and software to study human brain imaging with particular focus on the study of the disordered brain. This focus manifests in contributions to multiple areas of science including data-driven medical imaging analysis, functional connectivity, data fusion of multimodal imaging data, imaging genetics, and neuroinformatics among others. To list just a few of the contributions I am proud of:
• Established a 'first of its kind' multi-institutional neuroimaging and data science center (TReNDS)
• Among the first to develop centralized data sharing and neuroinformatics infrastructure. This include implementation of fully automated 'cloud-based' neuroimaging facility at the center for advanced brain imaging (CABI; http://cabi.gsu.edu) which I direct, fueled by COINS (http://coins.trendscenter.org) and BrainForge (https://brainforge.trendscenter.org/) automated
analysis tools developed at GSU.
• Published over 1000 peer-reviewed journal papers published, and over 1000 conference papers.
• Developed a brain network estimation approach (group ICA), which has since become a widely
used tool in the field and is a large part of various national and international projects including the NIH funded human connectome project. Our toolbox, GIFT, implements many dozens of approaches and algorithms and is among the most widely used in the field)
• One of the first applications of a multivariate network approach to study mental illness including schizophrenia and bipolar disorder.
• First to develop joint blind source separation approaches to identify multimodal markers of brain disorders (including multimodal imaging and imaging- genomics)
• An early strong advocate for single-subject classification approaches including one of the first classification studies in schizophrenia, multiple review articles, and a highly cited special issue in neuroimage on the topic.
• First to develop techniques to study whole brain dynamic connectivity (his introduction of techniques for this essentially launched a whole new direction in the field and was highlighted by the National Institutes of Mental Health Director in the 'best of the year' list)
• Introduced a family of multivariate analyses of imaging (epi)genomic data and identified some key links between structural/functional brain networks and genomic patterns.
• One of the first to introduce and validate deep learning approaches for neuroimaging, including a high turnout symposium and educational session at the OHBM meeting on the topic.
• Developed methods to study spatial dynamics (spatially varying nodes) within network/multivariate modeling approaches. Prior to this point spatial dynamics was essentially ignored.
• Broad and deep application of the above approaches to many areas including mental and neurological disorders, normative development, healthy aging, and more.
• First to introduce algorithms and to develop tools (e.g., COINSTAC, http:// https://coinstac.trendscenter.org/) for federated learning from large scale brain imaging.
• Prolific software developer including the widely used GIFT (group ICA of fMRI and EEG) & FIT (data fusion) software packages, the COINS neuroinformatics tools in use by dozens of imaging facilities around the world, the COINSTAC federated analysis platform, and the BrainForge automated neuroimaging portal (developed at GSU).
• Discovery of deficiencies in communication efficiency between the 'default mode network' in the brain and other networks in patients with schizophrenia. Follow up studies are evaluating cross- domain entropy approaches to try to improve upon existing models of schizophrenia.
• Active in tech transfer work, including multiple patents and small business grants for predicting medication class response to antidepressants or mood stabilizers in emerging youth and for scaling up of cloud-based neuroinformatics tools for data capture, management, and sharing.
• First to show deep learning methods can outperform standard machine learning in terms of their prediction accuracy, while also be used in a way that can provide interpretable output, moving beyond black box modeling approach.
• Established a 'first of its kind' multi-institutional neuroimaging and data science center (TReNDS)
• Among the first to develop centralized data sharing and neuroinformatics infrastructure. This include implementation of fully automated 'cloud-based' neuroimaging facility at the center for advanced brain imaging (CABI; http://cabi.gsu.edu) which I direct, fueled by COINS (http://coins.trendscenter.org) and BrainForge (https://brainforge.trendscenter.org/) automated
analysis tools developed at GSU.
• Published over 1000 peer-reviewed journal papers published, and over 1000 conference papers.
• Developed a brain network estimation approach (group ICA), which has since become a widely
used tool in the field and is a large part of various national and international projects including the NIH funded human connectome project. Our toolbox, GIFT, implements many dozens of approaches and algorithms and is among the most widely used in the field)
• One of the first applications of a multivariate network approach to study mental illness including schizophrenia and bipolar disorder.
• First to develop joint blind source separation approaches to identify multimodal markers of brain disorders (including multimodal imaging and imaging- genomics)
• An early strong advocate for single-subject classification approaches including one of the first classification studies in schizophrenia, multiple review articles, and a highly cited special issue in neuroimage on the topic.
• First to develop techniques to study whole brain dynamic connectivity (his introduction of techniques for this essentially launched a whole new direction in the field and was highlighted by the National Institutes of Mental Health Director in the 'best of the year' list)
• Introduced a family of multivariate analyses of imaging (epi)genomic data and identified some key links between structural/functional brain networks and genomic patterns.
• One of the first to introduce and validate deep learning approaches for neuroimaging, including a high turnout symposium and educational session at the OHBM meeting on the topic.
• Developed methods to study spatial dynamics (spatially varying nodes) within network/multivariate modeling approaches. Prior to this point spatial dynamics was essentially ignored.
• Broad and deep application of the above approaches to many areas including mental and neurological disorders, normative development, healthy aging, and more.
• First to introduce algorithms and to develop tools (e.g., COINSTAC, http:// https://coinstac.trendscenter.org/) for federated learning from large scale brain imaging.
• Prolific software developer including the widely used GIFT (group ICA of fMRI and EEG) & FIT (data fusion) software packages, the COINS neuroinformatics tools in use by dozens of imaging facilities around the world, the COINSTAC federated analysis platform, and the BrainForge automated neuroimaging portal (developed at GSU).
• Discovery of deficiencies in communication efficiency between the 'default mode network' in the brain and other networks in patients with schizophrenia. Follow up studies are evaluating cross- domain entropy approaches to try to improve upon existing models of schizophrenia.
• Active in tech transfer work, including multiple patents and small business grants for predicting medication class response to antidepressants or mood stabilizers in emerging youth and for scaling up of cloud-based neuroinformatics tools for data capture, management, and sharing.
• First to show deep learning methods can outperform standard machine learning in terms of their prediction accuracy, while also be used in a way that can provide interpretable output, moving beyond black box modeling approach.
Vaibhavi Itkyal
Neuroscience
Entrance Year: 2021
Topic: Multimodal Data Fusion, Data Science, Neuroimaging
