Efficient Preconditioning in Joint Total Variation Regularized Parallel MRI Reconstruction
Parallel magnetic resonance imaging (pMRI) is a useful technique to aid clinical diagnosis. In this paper, we develop an accelerated algorithm for joint total variation (JTV) regularized calibrationless Parallel MR image reconstruction. The algorithm mini
- PDF / 1,384,868 Bytes
- 8 Pages / 439.363 x 666.131 pts Page_size
- 70 Downloads / 171 Views
2
Department of Computer Science and Engineering, University of Texas at Arlington, Arlington TX 76019, USA Department of Radiology, New York University, New York, NY 10016, USA [email protected]
Abstract. Parallel magnetic resonance imaging (pMRI) is a useful technique to aid clinical diagnosis. In this paper, we develop an accelerated algorithm for joint total variation (JTV) regularized calibrationless Parallel MR image reconstruction. The algorithm minimizes a linear combination of least squares data fitting term and the joint total variation regularization. This model has been demonstrated as a very powerful tool for parallel MRI reconstruction. The proposed algorithm is based on the iteratively reweighted least squares (IRLS) framework, which converges exponentially fast. It is further accelerated by preconditioned conjugate gradient method with a well-designed preconditioner. Numerous experiments demonstrate the superior performance of the proposed algorithm for parallel MRI reconstruction in terms of both accuracy and efficiency.
1
Introduction
Parallel MR imaging is a powerful method that uses multiple receiver coils for reducing scanning time in MRI[5,6]. Based on the way in utilizing the sensitivity information and local kernel in k-space, these methods are classified broadly into two main types. Reconstruction techniques such as SENSE[11] and CSSENSE[8] expect accurate estimation of reception profiles from each coil element to optimally reconstruct undersampled MR image. However, it is often very difficult to accurately and robustly measure the sensitivities and even small errors can result in inconsistencies that lead to visible artifacts in the image. These disadvantages therefore motivate the other type of methods, termed auto-calibrating methods, e.g. GRAPPA[4] and SPIRiT[9], that derive sensitivity information from autocalibration signals (ACSs) and thus avoid side effects brought by the difficult and inaccurate sensitivity map estimation. However, it is often limiting or totally infeasible to acquire sufficient ACSs. For example, for non-Cartesian imaging, ACS acquisition requires much longer time and can probably lead to artifacts due to off-resonance. To overcome these shortcomings, several calibrationless methods have been proposed recently, e.g. CaLMMRI[10], FISTA JTV[2] and SAKE[15].
This work was partially supported by U.S. NSF IIS-1423056, CMMI-1434401, CNS-1405985. Corresponding author.
c Springer International Publishing Switzerland 2015 N. Navab et al. (Eds.): MICCAI 2015, Part II, LNCS 9350, pp. 563–570, 2015. DOI: 10.1007/978-3-319-24571-3_67
564
Z. Xu et al.
Among these methods, joint total variation (JTV) model has been demonstrated as a powerful tool for calibrationless parallel MR image reconstruction. The JTV model is designed based on the observation of the gradient sparsity of each coil image and the cross-channel similarity of parallel MR images. Previous attempt to solve this model is shown in [2] which is based on FISTA JTV algorithms. The numerical experiments exhibit its e
Data Loading...
