A Novel Approach Of Obtaining Optimal Solution For Iterative Self-Consistent Parallel Imaging Reconstruction

As Magnetic Resonance Imaging (MRI) is gaining popularity in the field of medical science, there are many ongoing researches to increase it’s accuracy and efficiency. In this work, the focus has been given to an already established parallel MRI reconstruction method named Iterative SelfConsistent Parallel Imaging Reconstruction: SPIRiT. SPIRiT uses coil-by-coil parallel reconstruction method and then applies sum-of-square (SOS) to produce the final MR image. The SOS recombination technique proved to be amplifying noise as well as it based on the assumption that parallel coil sensitivities are uniform. In this paper a convex optimization scheme has been implemented which not only can overcome the shortcomings of SOS recombination technique but also can provide a true optimal solution for SPIRiT. The SOS recombination of original SPIRiT has been replaced by a convex optimization formula and hen the upgraded SPIRiT method has been compared with original SPIRiT for proposed method’s legitimacy. It has also been shown that this research also has impacts on the scanning time. The comparison results cover the aspects of qualitative, quantitative and noise-based analysis to illustrate the efficiency of the proposed formula.