Research Abstracts Online
2008 - March 2009
University of Minnesota Twin Cities
Institute of Technology
Department of Biomedical Engineering
PI: Theoden Netoff
Data Processing for Fluorescence Optical Imaging of Neuronal Activity
These researchers take large datasets and filter and perform independent component analysis on the data using MATLAB scripts. This analysis requires a considerable amount of memory and number crunching and is greatly facilitated by using a cluster of machines to distribute the computation. This also allows the researchers to analyze data while simultaneously acquiring new datasets and to get feedback from the analysis during the experiment. This would not be possible with the computational power of standard desktop computers.
The researchers are also running simulations of neuronal networks to better understand epilepsy. They do this by running parameter searches on their virtual networks to test for transitions in behaviors on the continuum from normal to seizing.
Another aspect of this project is seizure prediction using support vector machines. The group is processing 15 TB of electroencephalograph data obtained from the Mayo Clinic in Rochester, Minnesota to generate algorithms that can distinguish an EEG under normal conditions to one taken immediately before a seizure.
Amarpreet S. Bains, Supercomputing Institute Undergraduate Intern
Bryce Beverlin II, Graduate Student
Dongyon Kang, Undergraduate Student
Oscar Miranda-Dominguez, Graduate Student
Yun Sang Park, Graduate Student
Tyler Stigen, Graduate Student
Chris Warren, Research Associate