Dr. Joshua Vogelstein

SLI Title

Assistant Professor, Department of Biomedical Engineering and the Institute for Computational Medicine


Joshua T. Vogelstein received a B.S degree from the Department of Biomedical Engineering (BME) at Washington University in St. Louis, MO in 2002, a M.S. degree from the Department of Applied Mathematics & Statistics (AMS) at Johns Hopkins University (JHU) in Baltimore, MD in 2009, and a Ph.D. degree from the Department of Neuroscience at JHU in 2009. He was a Postdoctoral Fellow in AMS@JHU from 2009 until 2011, at which time he was appointed an Assistant Research Scientist, and became a member of the Institute for Data Intensive Science and Engineering. He spent 2 years at Information Initiative at Duke, before coming home to his current appointment as Assistant Professor in BME@JHU, and core faculty in both the Institute for Computational Medicine and the Center for Imaging Science. He married his kindergarten sweetheart in the summer of 2014. His research interests primarily include computational statistics, focusing on big data, wide data, and icky data, especially connectomics. His research has been featured in a number of prominent scientific and engineering journals and conferences including Annals of Applied Statistics, IEEE PAMI, NIPS, SIAM Journal of Matrix Analysis and Applications, Science Translational Medicine, Nature Methods, and Science.

My research passion lies in the development of inference techniques for scientific discovery from large and complex datasets, typically relating measurements of brain properties (e.g., brain imaging) to mental properties (e.g., aptitude, cognition, perception, memory, etc.). Of primal consideration for our group is that these techniques are useful in solving important scientific questions and social problems. Our unique contributions follows from the juxtaposition our collective domain knowledge (enabling important applied questions to be asked), computational aptitude (allowing the tools that enable one to obtain answers from terabytes of data to be built), and statistical insight (clarifying the extent to which to trust the answers). All projects are motivated by scientific questions, and result in open source code available to the greater scientific community, as well as applications of the methods on state-of-the-art neuroscientific datasets. For all projects, we primarily search for students who are excited and fun to work with, secondarily, it would be helpful if you had some math/stat skills, some programming skills, and some interest in solving real problems. Below, are a few example projects.

  • Human connectomics: Even the simplest questions about human connectomes (brain-graphs) - such what is the mean connectome, and are these two populations of connectomes different - remain unanswered. In part, this is due to a lack of theoretically justified tools with resulting scalable code and appropriate data to answer them. We have developed fundamental theory (e.g., Tang14a), and scalable code (FlashGraph), that will now enable us to rigorously answer these questions. We are looking for students who would like to extend and apply these methods to extant collected and pre-processed human connectomes (available here), to get some quantitative answers. A video describing some of the related data can be found here.
  • Synaptic diversity: All brains are composed of neurons, and the connections between them, called synapses. All brain function is therefore at least partially determined by these synapses, and yet, we are woefully clueless with regard to their diversity and functional properties. We do know that each synapse is a complex composition of hundreds of proteins, all acting in concert together (read O'Rourke12a for a nice review). We have the data, as well as the methodological tools to enable exploring this diversity. We are looking for students who would like to extend and apply these methods to the extant collected and pre-processed data sets (available here), to reveal previously unknown but fundamental principles of neural computation. Videos describing these datasets can be found here.
  • Electron microscopy connectomics: Microcircuits of the brain enable computational capabilities far exceeding our greatest modern supercomputers, yet they fit in little skulls weighing less than a grapefruit. It is widely believe that cracking the neural circuitry code will reveal mysteries of the brain. To do so, however, we need petascale computer vision methods, such as three-dimensional scene parsing, as well as statistical tools to analyze these large and complex datasets. We are looking for students who would like to extend and apply these methods to the extant collected and pre-processed data sets (available here). Videos describing these datasets can be found here and here.
  • Connectome coding: Understanding the relationship between brain properties (for example, connectivity) and mental properties (for example, memories) remains one of the greatest challenges in the 21st century. It is also an example of a class of cutting-edge machine learning problems, sometimes called multi-modal learning. We have already developed some tools (e.g., JOFC) and have pre-processed some datasets (e.g., Adelstein et al, 2001), we are looking for students to extend and apply these methods to uncover, for example, relationships between brain networks and personality types.

Affiliated Research