INTRODUCTION

Recently, we introduced a novel MR pulse sequence for fast and yet quiet 3D radial, multi-gradient echo MR imaging. The method was termed Looping Star¹, reflecting its circular k‐space traverse and the 3D radial image encoding. Looping Star was demonstrated for quiet, submillimeter T2* and QSM imaging, and quiet BOLD fMRI^1-3.
The Looping Star pulse sequence is based on the 3D radial Rotating Ultra‐Fast Imaging Sequence (RUFIS)⁴, extended by a time‐multiplexed gradient‐refocusing mechanism. Accordingly, Looping Star captures an initial free-induction-decay (FID) image followed by gradient echo (GRE) images at equidistant echo times. In the original implementation of Looping Star, gradient refocusing was affected by overlapping echo-in and echo-out signals. The associated artifact needed correction either via filtering (halving resolution), or an extra acquisition with alternating RF phase cycling (doubling scan time).
Here now we present an improvement of the original Looping Star method, which eliminates the echo in/out overlap and associated artifacts, allowing faster scanning and/or higher resolution.

METHODS

RUFIS⁴ (Fig. 1 left) can be considered the simplest possible pulse sequence, consisting of 1) a constant readout gradient (black line) with only minor directional gradient updates (black dot) in between repetitions (hence silent) and 2) ultra-short, hard-pulse RF excitation (red) such that the excitation bandwidth encompasses the full imaging bandwidth. With the readout gradient active the whole time, image encoding starts effectively at the moment of RF excitation resulting in center-out radial sampling with TE = 0 (hence often referred to as Zero TE).
In standard Looping Star (Fig. 1 middle), the RUFIS sequence is divided into an RF excitation phase (left) and a gradient refocusing phase (right) using a self-refocusing k-space trajectory. In the example shown, six FID signals are excited and encoded using 3D radial spokes with a relative phase of 60deg. By repeating the same spokes during the refocusing phase, all excited FID signals are looping along a hexagonal trajectory back into center of k-space forming a gradient echo. During the refocusing, the signal of the current spoke (echo-out) overlaps with the next refocusing spoke (echo-in), resulting in artifacts. Previously, the echo in/out overlap problem was addressed via filtering (halving resolution), or RF-phase cycling (doubling scan time).
The new Looping Star scheme (Fig. 1 right) addresses the echo in/out problem by exciting only every second spoke during the exciting phase. As a consequence, the echo in/out signals are now separated such as to not overlap during the refocusing phase. [In the example shown, only three FID center-out signals (1, 3, 5) are excited. During the refocusing, (6 & 7) capture the echo-in/out signal of 1; (8 & 9) capture the echo-in/out of 3; (10 & 11) capture the echo-in/out signal of 5.]
For 3D spatial encoding, the loops and its constituent spokes must be sampled along different directions. The orientation of the loops was numerically optimized for uniform sampling density. The new Looping Star pulse sequence was implemented and tested on a 3T GE MR750w scanner using an 8-channel brain coil (GE Healthcare, Chicago, IL). Image reconstruction was implemented offline in Matlab (Mathworks, Natick, MA) using standard 3D nearest-neighbor gridding⁵.

RESULTS

Figure 2 shows phantom results comparing original vs. new Looping Star (FOV=192mm, res=1mm, BW=±31.25kHz, FA=2, N_SpkPerLoop=8, N_loop=4, ΔTE=14.8ms, LAeq=68.4dB(A)). The overlapping echo in/out signals generate spatially-incoherent, noise-like background, which can be corrected via filtering (reducing resolution), or RF phase cycling (doubling scan time). The new Looping Star method intrinsically avoids echo in/out overlap resulting in a sharp image with clean background while still maintaining resolution and scan time.
Figure 3 illustrates corresponding in-vivo brain images for the new Looping Star method (same parameters as above), demonstrating 3D high-resolution, quiet imaging, providing an FID image (left) and three equidistant GRE images with ΔTE=14.8ms.
Figure 4 compares a quiet fMRI protocol (FOV=19.2mm, res=3mm, BW=±31.25kHz, FA=2, N_SpkPerLoop=32, N_loop=2, ΔTE=26.88ms, LAeq=66.9dB(A)) for conventional Looping Star with phase cycling and the new Looping Star method (each fully sampled and 4x accelerated). The new Looping Star method demonstrates similar resolution and image quality but 2x shorter scan time.

DISCUSSION

Looping Star is a new concept for acquiring FID and GRE images in an efficient and yet quiet manner. The described improvement eliminates echo in/out signal overlap and thereby increases scan efficiency in terms of resolution and/or scan time. The new Looping Star method samples twice as many spokes for the GRE image (echo in/out) compared to the FID image (echo out only), which is a favorable tradeoff given the higher prevalence of GRE information. Fewer excitations also allows higher flip angles and hence higher SNR.
All results shown were obtained using a simplistic, nearest-neighbor 3D gridding reconstruction⁵. Given its 3D radial center-out sampling significant scan acceleration is anticipated by using more advanced image reconstruction methods^5,6.
From an application perspective, we expect Looping Star to be particularly useful for T2* BOLD fMRI, where its quiet imaging performance enables new investigations (e.g. auditory, sleep, resting state, …) and/or scanning patients who otherwise would not tolerate a loud MR scanner environment.

Acknowledgements

No acknowledgement found.