Automatic Gating of Flow Cytometry Data
With Gyemin Lee, Ph.D. Student,
University of Michigan
Faculty Host: Venkatesh Saligrama
Refreshments will be served outside Room 339 at 2:45 p.m.
Abstract: Flow cytometry is a technique for rapidly quantifying physical and chemical properties of large numbers of cells. In clinical applications, flow cytometry data must be manually "gated" to identify cell populations of interest. Because multiple iterative gates are often required to identify and characterize these populations, several researchers have investigated statistical methods for automating this process.
Gyemin Lee will first present an unsupervised learning technique based on multivariate mixture models. Since measurements from a flow cytometer are often censored and truncated, standard model-fitting algorithms can cause biases and lead to poor gating results. He and his research team propose novel EM algorithms for fitting multivariate Gaussian mixture models to data that is truncated, censored, or truncated and censored.
In the second part, he will view the problem as one of transfer learning. Combined with the low-density separation principle, this method can leverage existing datasets previously gated by experts while accounting for biological variation to automatically gate a new flow cytometry dataset.
Lee will demonstrate these techniques on clinical flow cytometry data and evaluate their effectiveness.
About the Speaker: Gyemin Lee received his B.S. degree in electrical engineering from Seoul National University, Seoul, Korea, in 2001, and his M.S. degree in electrical engineering from University of Michigan, Ann Arbor, in 2007, where he is currently pursuing his Ph.D. in Electrical Engineering: Systems under the supervision of Professor Clayton Scott. His research interests include machine learning, optimization, and biomedical applications. |
