Meeting Banner
Abstract #2609

Empirical Investigation of the Gardner Transform as a Sparsifying Transform for the Analysis of a New Class of Signals Using Compressed Sensing

Jordan Woehr1, Michael Smith1, 2

1Electrical and Computer Engineering, University of Calgary, Calgary, Alberta, Canada; 2Radiology, University of Calgary, Calgary, Alberta, Canada

Compressed sensing (CS) applied to under-sampled <i>k</i>-space is used in magnetic resonance to decrease 2D and 3D imaging times while maintaining image resolution. We present the Gardner transform (GT) as a potential sparsifying transform for use with CS with under-sampled or truncated signals in a non-<i>k</i>-space 4th dimension, e.g. time or frequency. New classes of signals can be sparsified with the GT such as sums of exponentials, Lorentzians, etc. We discuss the practical issues associated with GT-CS reconstruction of a simulated signal assumed to have two exponential components, MTTGM and MTTWM, and an ideal GT of two Dirac delta functions.

Keywords

acquisition almost applicable application applying arise assumed attempts averaging basis behind black block blue boundaries central characteristics class column combined combining component components compressed computer consistency constraint containing continuity core criterion decay deconvolution decrease delta derived despite determining discontinuities discrete domain domains either electrical empirical engineering equation examined explained exponential exponentials extrapolation fails filtered filtering frequency function functions generated generating generation gray half hypothetical ideal identified illustrating improved intensity introduced investigation issues kernel known lack lead leads linear lost magnitudes minimize missing modeling narrow noise norm norms note operation partial peak peaks phys practical preconditioning preliminary presumed proportional propose radiology random rather real reason reconstruction reconstructions recover recovered reducing removes removing representation requires resolution respectively response retaining sampling sensing series several shape short side simulation simulations since smaller smith software solution space sparse sparseness spectral spots stabilizing substitution sums tall theory together transform truncated unless useful volume wavelet white wide widening