Compressed Sensing Reconstruction Improves Variable Density Spiral Functional MRI

Introduction: Recent approaches to spiral imaging using variable density (VD) trajectories [1,2] have demonstrated the ability to decrease the data acquisition window for an equivalent image matrix size, with a subsequently improved fMRI sensitivity attributed to a higher time course sampling rate. In separate research, the application of Compressed Sensing (CS) reconstruction algorithms to sparse k-space data, such as VD spiral, has been shown to decrease the aliasing artifact associated with sub-Nyquist data sampling [3]. However, the use of CS reconstructed VD (CS-VD) spiral data has not been demonstrated in terms of whether its use improves sensitivity to BOLD during fMRI acquisitions, relative to conventional reconstruction methods. In the current work we demonstrate that the use of CS-VD spiral data acquired during high field (4-T) fMRI minimizes the aliasing artifact inherent to sparse k-space acquisitions that can result in additional signal fluctuations in the time course data, thereby improving the apparent sensitivity to cortical activity relative to the same images reconstructed without CS.