· · · Repository hosted by TU Delft Library

Fast reconstruction is crucial for the implementation of CS-SENSE onclinical scanners. Thus, improvements of the reconstruction speed are desirable, both in terms of algorithms with improved convergence and parallel implementation. In this work, we propose a modified CS-SENSE reconstruction method based on the Nesterov’s optimal gradientscheme, which is less sensitive to inaccuracies in the coil sensitivity estimation and has an improved convergence speed.

In this work, we investigate the influence of the sampling pattern on the convergence behaviour of {1-regularized SENSE reconstruction at different reduction factors. In other words, we try to answer the question what improvement can CS-SENSE provide over {1-denoised SENSE.

In this work, we present a CS reconstruction based on statistical non-local self-similarity filtering (STAINLeSS), in which the parameters are entirely determined by the noise estimation in the receive channels obtained from a standard noise measurement. The method achieves improved image quality compared to wavelet based CS reconstruction in particular in SENSE based multi-coil reconstruction due to itsadaptivity to spatially varying noise. The proposed method providesimproved robustness due to the lack of free parameters which is crucial for the clinical applicability of CS.