Efficient MRI Reconstruction Using a Hybrid Framework for Integrating Stepwise Bayesian Restoration and Neural Network Models in a Memory Based Priors System

• Authors: D. A. Karras
• Venue: HAIS '08 Proceedings of the 3rd international workshop on Hybrid Artificial Intelligence Systems
• Year: 2008

The goal of this paper is to present a new image restoration methodology for extracting Magnetic Resonance Images (MRI) from reduced scans in k-space. The proposed approach considers the combined use of Support Vector Machine (SVM) models in a memory based Bayesian priors system and a novel Bayesian restoration, in the problem of MR image reconstruction from sparsely sampled k-space, following several different sampling schemes, including spiral and radial. Effective solutions to this problem are indispensable especially when dealing with MRI of dynamic phenomena since then, rapid sampling in k-space is required. The goal in such a case is to make measurement time smaller by minimizing scanning trajectories. It is suggested here that significant improvements could be achieved, concerning quality of the extracted image, applying to the k-space data a novel Bayesian restoration method based on progressively approximating the unknown image, involving a neural model based priors system with memory specifically aimed at minimizing first order image derivatives reconstruction errors. More specifically, it is demonstrated that SVM neural network techniques could construct efficient memory based Bayesian priors and introduce them in the procedure of a novel stepwise Bayesian restoration. These Priors are independent of specific image properties and probability distributions. They are based on training SVM neural filters to estimate the missing samples of complex k-space and thus, to improve k-space information capacity. Such a neural filter based Bayesian priors system with memory is integrated to the maximum likelihood procedure involved in the Bayesian reconstruction and aims at minimizing image derivatives and image reconstruction errors. It is found that the proposed methodology leads to enhanced image extraction results favorably compared to the ones obtained by the Integrated Bayesian MRI reconstruction involving simple SVM models priors as well as memory based priors minimizing image reconstruction errors instead of its derivatives errors, by the traditional Bayesian MRI reconstruction as well as by the pure Neural filter based reconstruction approach.