Compressed sensing MRI combined with SENSE in partial k-space.

Compressed sensing (CS), parallel imaging and partial Fourier (PF) acquisition are all effective methods to reduce k-space sampling and therefore accelerate MR acquisition. The combined use of these methods gives us more options to balance the needs for scan speed and image quality. We conducted simulations on full k-space data to demonstrate the potential use of combining CS-SENSE with PF acquisition in anatomical MRIs of the human brain. To test the accelerated acquisition of high-resolution T1-weighted images of brain, we modified a 3D FSPGR sequence on a GE 3T scanner to implement different undersampling schemes based on CS, including partial Fourier CS-SENSE. Partially sampled k-space data were acquired and then reconstructed to brain images. CS-SENSE combined with PF sampling is able to provide better reconstructed images than CS only, or than CS-SENSE without PF for the same total acceleration. Combining PF sampling with CS-SENSE enables us to further accelerate image acquisition or improve image quality while holding the acceleration rate constant.