Accelerating Parallel Magnetic Resonance Imaging Using p-Thresholding Based Compressed-Sensing

Abstract

Conjugate gradient-based SENSE (CG-SENSE) and compressed-sensing (CS) are well-established techniques to accelerate magnetic resonance imaging (MRI) data acquisition. CG-SENSE is an iterative parallel MRI (pMRI) technique, used for the reconstruction of unaliased MR images from the under-sampled arbitrary k-space trajectories (Cartesian and non-Cartesian). Whereas CS is a promising technique that requires fewer random samples in the k-space to speed up the data acquisition process for MR image reconstruction. In the recent past, further acceleration in MR data acquisitions has been achieved using pMRI and CS jointly. In this paper, a novel method is proposed which sequentially combines CG-SENSE with p-thresholding based CS to achieve higher acceleration factors without compromising the quality of image reconstruction. In the proposed method, CG-SENSE and p-thresholding based CS reconstructions are sequentially combined to recover aliased free images from highly under-sampled k-space data. The performance of the proposed method is evaluated for arbitrary k-space Cartesian and radial trajectories. The reconstruction results are compared with conventional methods, e.g., CG-SENSE and $$\ell_{1}$$ -SPIR-iT. Several experiments are performed using simulated phantom and in vivo datasets. The reconstruction quality of the proposed method is evaluated in terms of artifact power (AP), peak signal-to-noise ratio (PSNR) and root mean square error (RMSE). The experimental results show that the proposed method outperforms the CG-SENSE and $$\ell_{1}$$ -SPIR-iT by achieving superior image reconstruction quality.