Data Acquisition Strategies for Reducing Eddy-Current and Transient Oscillation Artifacts in Balanced Steady-State Free Precession
Hyun-Soo Lee1, Seung Hong Choi2, and Sung-Hong Park1

1Department of Bio and Brain Engineering, Korea Advanced Institute of Science and Technology, Daejeon, Korea, Republic of, 2Department of Radiology, Seoul National University College of Medicine, Seoul, Korea, Republic of

Synopsis

The quality of balanced steady-state free precession is vulnerable to eddy-currents and transient oscillations. However, the conventional centric phase-encoding (PE) scheme makes these artifacts severe, thus needs additional compensation strategies. In this study, we propose an improved PE scheme where k-space is encoded from center to periphery in a group-wise manner (PE-grouping). This reduces related artifacts by preventing big jumps in k-space along PE direction. Also proposed were various averaging strategies that could further eliminate the residual artifacts by averaging two full images acquired not only with the PE-grouping, but also with the conventional centric and pairing schemes.

Introduction

For physiological MR imaging (i.e., diffusion, perfusion, magnetization transfer imaging), balanced steady-state free precession (bSSFP) is frequently used due to its excellent signal-to-noise ratio (SNR) and short scan time. For data acquisition, a centric phase-encoding (PE) order is preferred to a linear order in order to maximize prepared magnetization contrast. However, the conventional centric PE scheme induces severe signal oscillations due to non-constant eddy-currents. In this study, we propose (i) a pseudo-centric PE-grouping (PE-grouping) which is a hybrid of the conventional centric and linear orders for optimal contrast and minimal signal fluctuations. Also, we propose (ii) four averaging strategies which further reduce the residual eddy-current and transient oscillation artifacts by averaging two full images so that oppositely-oscillating signals of two images can be canceled out. The proposed strategies markedly removed eddy-current and transient oscillation artifacts for both phantom and in vivo while maintaining the SNR and temporal resolution/scan of the conventional centric scheme, which is better than the pairing [1] and comparable to existing double averaging scheme (dAVE) [2].

Theory

In principle, eddy-current-induced dephasing of transverse magnetization is proportional to the amplitude of gradients but with opposite polarity [3]. With the conventional centric PE scheme, the amplitude and polarity of PE gradients change every repetition time (TR), which induces time-varying eddy-current dephasing resulting in severe spatial phase offsets. Since bSSFP is robust to constant or slowly-changing spatial phase errors [4], we propose PE-grouping in which PE lines with a certain number N are grouped and the PE-groups are linearly encoded from k-space center to edge (Fig. 1d). This method is expected to reduce eddy-current-induced signal oscillations because the number of jumps in k-space are reduced. Also, the signal oscillations including transient oscillation can be canceled out by averaging two signals that have opposite oscillation patterns. The phases of oscillations can be adjusted by changing RF phases of PE line of bSSFP (i.e., phase cycling angle). Four average schemes can be differentiated by the way two images have opposite RF phases for the same PE lines: “dummy average” for different number of dummy scans, “backward average” for opposite PE order within groups, “up-down average” for different start point of encoding as upper or lower-half of k-space, and “up-down-backward” as the combination of preceding two average schemes.

Data Acquisition

All experiments were performed on a Siemens 3T Trio system (Siemens Medical Solution, Erlangen, Germany) on a doped water phantom (measured =290/230 ms) and a volunteer. The half alpha preparation was performed with no dummy scans and measurements were performed five times repeatedly for each scan, in order to investigate the differences between the transient and the steady states. A delay time of 10 s was applied before each different acquisition (not between the five dynamic scans), considering longitudinal relaxation time.

Results

In Fig. 2a, the oscillation peak of the conventional centric scheme was reduced with PE-grouping of both N=3 and 4, which was lower than that of the pairing (N=2). Up-down average further reduced the peaks, even lower than dAVE. Also, the number of peaks increased with N of PE-grouping, while their amplitudes decreased (Fig. 2b-c). The two signals from the conventional centric scheme with zero (black line) and one (gray line) dummy scans oscillated in opposite phases to each other; thus the transient oscillations were markedly reduced after complex averaging (Fig. 3). For phantom, the artifacts of the conventional centric scheme (indicated by arrowheads) were reduced with PE-grouping (N=4) for both inner and outer parts of the phantom, while the background artifact increased with the pairing; furthermore, the four average schemes clearly removed artifacts comparable to dAVE (Fig. 4, top). In vivo results (Fig.4, bottom) confirmed that the proposed strategies clearly removed artifacts compared to the existing methods (N=1 and N=2) for both transient (top) and steady-states (bottom).

Discussion and Conclusion

The PE-grouping considerably reduced eddy-current-induced signal oscillations and proposed average schemes further eliminated artifacts including transient oscillations, which is beyond or comparable to existing compensation schemes. For the choice of N with PE-grouping, there are some tradeoffs between stability and contrast. Signal oscillations are reduced with greater N independent of off-resonance condition (Fig. 2c), while there would be bigger heterogeneity in contrast for physiological imaging due to the prolonged delay time between PE-groups near k-space center. However, proposed average schemes can reduce artifacts for small N, where the heterogeneity is negligible. Therefore, the proposed schemes are expected to improve the image quality of physiological MRI with the bSSFP readout because of reduced artifacts from eddy-currents and transient oscillations, smaller number of dummy scans, and equal temporal resolution.

Acknowledgements

No acknowledgement found.

References

1. Bieri, Magn Reson Med 2005; 54:129-137. 2. Markl, Magn Reson Med 2005; 54:965-975. 3. Ahn, Magn Reson Med 1991; 17(1):149-163. 4. Foxall, Magn Reson Med 2002; 48:502-508.

Figures

Fig. 1. Schematic illustration of the four different PE strategies. a: Linear PE order. b: Centric PE order. c: Pairing. d: PE-grouping (N stands for the number of PE lines in each group). ky is a spatial frequency along the phase encoding direction, and each circle represents a PE line.

Fig. 2. Impact of the choice of N for PE-grouping on eddy-current-induced signal oscillations. a: Eddy-current-induced signal oscillations for multiple off-resonance conditions with various PE schemes. b: The number of oscillation peaks with PE-grouping for various Ns. c: Mean peak amplitude with PE-grouping for various Ns.

Fig. 3. Evolution of transverse magnetization with conventional centric PE scheme before (a) and after (b) dummy average on transient-state. The black and gray lines in a represent transverse magnetizations simulated with number of dummy scans of 0 and 1, respectively. The oscillations are significantly reduced after dummy average (b).

Fig. 4. 2D bSSFP imaging for phantom (top) and in vivo (bottom). a: Conventional centric. b: Pairing. c: PE-grouping (N=4). d-g: Four average schemes (N=4). h: dAVE. The bottom row shows enhanced windowing for visualizing background artifacts (phantom). Both transient (top) and steady-state (bottom) images are shown (in vivo).



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
4305