Amitava Chatterjee, professor at Jadavpur University of Calcuta-India

LISSI's Meeting room, Vitry-sur-Seine

Compressive Sensing for Medical Imaging Applications

Abstract:

Compressive Sensing (CS) is a method to capture and represent compressible signals at a rate significantly below the Nyquist rate. Compressive Sensing employs non-adaptive linear projections that preserve the structure of the signal. The signal is then reconstructed from these projections using an optimization process. This talk starts with a brief introduction to compressible signals. Then it discusses transform coding and inefficiencies associated with it. This is followed by discussions on the compressive sensing problem which describes how a stable measurement matrix can be designed, followed by the compressive sensing measurement process in general. Then the advantages and drawbacks of different norm based methods in solving signal reconstruction algorithms are discussed in detail. In this context, two fundamental premises of CS, i.e. sparsity and incoherence, are introduced and their significance and importance are discussed in detail.
The second part of the lecture gives a detail account of compressive sensing in medical imaging. The discussion starts with why CS is required in magnetic resonance imaging (MRI) and what are the basic limitations of the data acquisition process employed in traditional MRI. The importance of sparsity or compressibility of MR images is discussed in detail followed by introduction of MRI as a compressed sensing system. The importance of incoherence, in this context is highlighted. Then, how image reconstruction in CSMRI can be carried out is discussed in detail. The talk concludes with brief discussions about how compressive sensing can be utilized in solving other medical imaging problems e.g. rapid 3-D angiography and whole heart coronary imaging.