Independent Component Analysis for Functional Magnetic Resonance ImagingTülay Adali, University of Mayland, Baltimore County
Thursday September 12 2002, 15:30 room 133, building 321, IMM, DTU
NOTE: Room/building number might change.
AbstractFunctional magnetic resonance imaging (fMRI) provides the opportunity to study brain function non-invasively and is a powerful tool utilized in both the research and clinical arenas. The analysis of fMRI can be challenging due to low signal-to-noise ratio, high dimensionality of the data, and the difficulty of modeling the brain function. Independent component analysis (ICA) has recently been applied to the analysis of fMRI data and has demonstrated considerable promise primarily due to its intuitive nature and flexibility in characterizing the brain function. In this talk, I first introduce a framework for understanding and studying the application of ICA to fMRI data. The model includes a synthesis stage, enabling simulations of fMRI-like signals, and an analysis stage for tracking how various processing strategies and their interactions affect the signals. I then demonstrate the use of the model for optimizing the processing stages of fMRI analysis, and for extending ICA to make inferences about a group of subjects. I then demonstrate the application of ICA to the analysis of fMRI data as it is acquired, in the complex domain, instead of only using the magnitude da ta.
For the task, a complex infomax algorithm that uses an analytic complex nonlinearity is introduced and it is shown that, when compared to the traditional complex infomax approach, the new algorithm improves the shape of the performance surface, decreases computational complexity and convergence time, and is capable of more general approximations. I conclude by noting the promise of ICA to increase our understanding of the brain function.
BiographyTulay Adali received the B.S. degree from Middle East Technical University, Ankara, Turkey, in 1987 and the M.S. and Ph.D. degrees from North Carolina State University, Raleigh, in 1988 and 1992 respectively, all in electrical engineering. In 1992, she joined the Department of Electrical Engineering at the University of Maryland Baltimore County, Baltimore, where she currently is an associate professor.
She has worked in the organization of a number of international conference and workshops including the IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP) and the the IEEE International Workshop on Neural Networks for Signal Processing (NNSP). She has been the general co-chair for the NNSP workshops 2001-2003. She is currently the secretary for the IEEE Neural Networks for Signal Processing Technical Committee and is serving on the IEEE Signal Processing Society conference board. She is the author or co-author of three book chapters and more than 150 refereed articles in journals and conference proceedings, is an associate editor for the Journal of VLSI Signal Processing Systems and guest editor for special issues for the IEEE Transactions on Neural Networks and the VLSI Signal Processing Systems.
Her research interests are in the areas of neural computations, adaptive signal processing, estimation theory, and their applications in biomedical image analysis, channel equalization, time-series prediction, and optical communications.
Dr. Adali is the recipient of a 1997 National Science Foundation CAREER Award and the provost's research faculty fellowship for the 1997-1998 academic year.