Simultaneous multi-slice reconstruction by Regularized Nonlinear Inversion

TL;DR

This paper introduces a novel Regularized Nonlinear Inversion method for simultaneous multi-slice MRI reconstruction, eliminating the need for prior coil sensitivity estimates and enabling improved real-time imaging capabilities.

Contribution

The paper presents SMS-NLINV, a new nonlinear inversion approach for SMS MRI that does not require prior coil sensitivity information, enhancing flexibility and robustness in dynamic imaging scenarios.

Findings

Verified accuracy of SMS MRI sequence including slice distance and flip angle

Confirmed square-root-like SNR benefit of SMS over conventional methods

Demonstrated effectiveness of SMS-NLINV with phantom and in-vivo images

Abstract

Increasing imaging speed is of utmost importance in in-vivo magnetic resonance imaging (MRI). With simultaneous multi-slice (SMS) MRI we can simultaneously acquire several slices of an object, which allows for higher undersampling factors compared to single- or conventional multi-slice measurements by exploiting axial coil sensitivity information. In this thesis, we give a short introduction to the physical principles of MRI, cover the basics of a FLASH based SMS MRI sequence and perform tests to verify its accuracy: We check the fidelity of the slice distance as well as the flip angle and confirm the square-root-like signal-to-noise ratio benefit of SMS compared to conventional multi-slice experiments. A g-factor analysis is used to determine a favorable Cartesian undersampling scheme for multi-slice data. Common reconstruction strategies for SMS MRI make use of previously estimated coil sensitivities to solve a linear equation. Here, we propose a new method of SMS MRI based on Regularized Nonlinear Inversion (NLINV): SMS-NLINV. This method does not require a priori knowledge about the coil sensitivities and is attractive especially for real-time imaging where coil sensitivities may change due to motion or interactive changes to the slice position. We characterize the algorithm and present reconstructed phantom and in-vivo images.