Mehmet Akçakaya, Harvard Medical School, United States
Sparse Signal Recovery from Linear & Nonlinear Measurements with Applications in High-Resolution Cardiac MRI
Many signals, such as images and videos, can be represented in a sparse manner (i.e. with a few non-zero coefficients) in a transform domain. In this talk, we consider the sampling of sparse signals both with linear measurements (e.g. compressed sensing) and non-linear measurements (e.g. magnitude-only measurements), as well as applications in medical imaging. By making connections to information theory, coding theory and high-dimensional geometry, we study the number of samples required to reconstruct sparse signals in a variety of scenarios, including compressed sensing, sparse phase retrieval and joint estimation of sparse signals & measurement parameters. We then concentrate on the use of such sampling schemes in magnetic resonance imaging (MRI). We show that the use of image structure beyond sparsity improves the reconstruction quality for compressed sensing in a variety of high-resolution applications. Finally, we discuss how non-linear measurements can be incorporated in MRI acquisition & reconstruction for further speed-up.
Mehmet Akçakaya received his Bachelor's degree in Electrical Engineering in 2005 with great distinction from McGill University, Montreal. He received his Ph.D. degree from the School of Engineering and Applied Sciences, Harvard University in 2010. He is currently a faculty member at the Harvard Medical School. He is a Junior Fellow of the International Society of Magnetic Resonance in Medicine (ISMRM), and was an ISMRM I. I. Rabi Award finalist for his work in accelerated MRI. His research interests include sparse signal processing, information theory and MRI.
Invited by the Department of Electrical & Computer Engineering
200 University Avenue West
Waterloo, ON N2L 3G1