Compressed Sensing: from Cosmology to Magnetic Resonance Imaging

seminar default logo
Speaker :  
Dr. Jean-Luc Starck (CEA/Saclay, France)
Location :  
G. Lianis Amphitheater, FORTH
Date :  

Time : 

Video
Abstract :

Radio interferometry is a technique by which the signal of a plane wave from a distant source is collected simultaneously by multiple telescopes and later combined taking into account the difference in the path between the astronomical source and each radio telescope. The discrete nature of the signal has recently been interpreted by our team through the ``Compressed Sensing'’ (CS) acquisition theorem. The CS approach supports the idea of using a specific mechanism, called sparsity, in order to reconstruct 2D images from the measured data. In Magnetic Resonance Imaging (MRI), similar type of observations are being made – sampling the signal from many discrete directions - thus providing an opportunity for new types of MRI imaging acquisition and reconstruction techniques.

We will first briefly introduce the concept of CS and sparsity following by a demonstration on how the resolution of a radio-astronomy image can improved by a factor four compared to the current state-of-art. Then we will present novel MRI acquisition schemes developed for the NeuroSpin MRI instrument of CEA, which is about to achieve a strength of 11.7 Tesla later this year. Using such CS acquisition techniques, similar to those developed for our astrophysical reconstruction methods, allows superb resolution in the reconstructed image with a significant acceleration of the MRI acquisition time. Results using real MRI measurements will be presented.