Improving ktSENSE by Adaptively Selecting the Regularization Image
Xu D, King K, Liang Z
University of Illinois at Urbana-Champaign
The recently-proposed ktSENSE has shown very encouraging results with a factor of 5-8 improvement in imaging speed for several dynamic imaging applications. This paper presents an improvement on the method by providing adaptively-selected the regularization image for image reconstruction. Specifically, a variable-density, sequentially interleaved sampling pattern with references is proposed for data acquisition. This scheme provides sufficient auxiliary data to reconstruct high-resolution time-varying images using a generalized series model, which are then used as regularization images (adaptive to each time frame) to remove aliasing effects due to undersampling along both the time and k axes. Cardiac imaging experiments were carried out using the proposed method and compared with ktSENSE and other methods, which showed a substantially improved temporal resolution and less image artifacts.