Improving Dynamic Parallel MRI Reconstruction via a Kernel-Based Learning Technique

Abstract

As an important radiology technology, magnetic resonance imaging (MRI) has been widely used in clinical applications. However, its low imaging speed restricts some clinical applications such as dynamic imaging. Parallel MRI was proposed to accelerate imaging speed by undersampling k-space data and applied on dynamic imaging like cardiac imaging. Due to undersampled k-space data, noise is a problem in reconstructed MR images. We propose a nonlinear technique to improve a temporal parallel MRI reconstruction method. Experimental results show that the proposed nonlinear technique outperforms the traditional method.