Multi-Slice Imaging Using a Time-Dependent Phase Constraint Through Hadamard Excitation with SENSE Reconstruction (M-PHENSE)

Abstract

Image acquisition time is one of the most important considerations for magnetic resonance imaging (MRI). This is especially true for dynamic imaging (such as functional MRI or dynamic contrast enhanced MRI) which require multiple scans of the same volume over time. As such, increasing throughput has been a major focus of many advanced image reconstruction techniques. Many of the current techniques focus on an in-plane acceleration by reducing the amount of data acquired per scan. While these methods do increase data acquisition speed, they do not increase throughput proportionally to the data reduction. Moreover, by reducing the amount of data acquired, the signal-to-noise ratio (SNR) is also reduced. These particular shortcomings can be addressed with through-plane acceleration - imaging multiple volumes simultaneously. However, multi-slice parallel imaging is still susceptible to noise amplification due to non-ideal sensitivity profiles (i.e., g-factor). There is an additional decrease in SNR if the separation between excited bands is small. The purpose of this work is to present a novel reconstruction technique, M-PHENSE, which has been designed to reconcile these deficiencies. M-PHENSE is a multi-slice imaging technique that integrates Hadamard slice encoding for a through-plane acceleration. This method allows for a time-dependent, preliminary image reconstruction and creation of complex, time-dependent, phase-constrained sensitivity profiles. M-PHENSE achieves this by using a slice-selective UNFOLD technique to perform the preliminary reconstruction. Through simulations of the technique, M-PHENSE is shown to be nearly independent of slice separation, while maintaining SNR for the R=2 case. Techniques for incorporating a number of slice excitations greater than two are explored, and the limitations of M-PHENSE are discussed. Additionally, M-PHENSE is found to be adaptable with in-plane acceleration methods, establishing a two-dimensional acceleration. A combination with UNFOLD is presented increasing the reduction factor to four and maintaining minimal SNR loss. Overall, the M-PHENSE technique is found to be a robust reconstruction method with many advantages over the currently available procedures.