Meeting Banner
Abstract #4583

Investigation of Sparsifying Transforms for Compressed Sensing in MRI Reconstruction

Christopher Baker1, Kevin King2, Dong Liang1, Leslie Ying1

1Eletrical Engineering, University of Wisconsin at Milwaukee, Milwaukee, WI, USA; 2Global Applied Science Lab, GE Healthcare, Waukesha, WI, USA

In compressed sensing (CS) MRI reconstruction, the level of sparsity and incoherence achieved by the transform affects the under-sampling that can be performed. This work investigates contourlets and the discrete cosine transform (DCT) as sparsifying transforms for CS reconstruction and compares them with the widely used wavelet transform. Results show that the contourlet transform performs about the same as the wavelet, while the DCT on small image patches outperforms the wavelet in CS reconstruction of MR images. The observation suggests that use of a DCT on small image patches may improve the CS reconstruction quality.