k-T Sparse GROWL: A Fast & Accurate Algorithm for Highly Accelerated Dynamic Imaging

Feng Huang¹, Wei Lin¹, George Randy Duensing¹, Arne Reykowski¹

The combination of partially parallel imaging (PPI) and compressed sensing (CS) has shown great potential for dynamic MRI. In this work, a self-calibrated PPI technique GROWL is combined with CS in k-t space for fast and accurate reconstruction with highly accelerated dynamic imaging. The proposed method is called k-t sparse GROWL. By using golden angle radial trajectory, real time speech MRI with flexible temporal resolution can be achieved by using 16 radial projections for each time frame. The reconstruction for a low artifact image using a 25616 16 data set needs less than 10 seconds in Matlab.