Quantitative evaluation of Compressed Sensing in MRI: Application to 7T time-of-flight angiography

Compressed Sensing is a recently developed technique that enables the reconstruction of sparsely sampled MR data, thus enabling much faster image acquisitions than using conventional approaches which are limited by the Nyquist criterion. Even though this method has been shown to provide visually acceptable results in a number of applications, no quantitative evaluation of the quality of the image data reconstructed using this method have yet been provided. In this paper we evaluate Compressed Sensing reconstruction in terms of sensitivity to its intrinsic reconstruction parameters as well as to the acquisition strategy both with simulated and experimental time-of-flight angiography data acquired on a 7T clinical MR scanner. Results show that Compressed Sensing reconstruction outperforms traditional methods of undersampling such as simple zero-filling and provides data with acceptable quality at undersampling rates up to a factor of ten. Such an acquisition framework therefore appears to be a valid and promising approach for reducing data acquisition time.