Prospective frequency correction for TE-averaged semi-LASER
Chu-Yu Lee1, In-Young Choi1,2,3, Peter Adany1, and Phil Lee1,3

1Hoglund Brain Imaging Center, University of Kansas Medical Center, Kansas city, KS, United States, 2Department of Neurology, University of Kansas Medical Center, Kansas City, KS, United States, 3Department of Molecular & Integrative Physiology, University of Kansas Medical Center, Kansas City, KS, United States

Synopsis

Frequency drifts during MRS acquisition results in broad and distorted spectral lineshapes, a reduced SNR and quantification errors. The consequence of frequency drifts is particularly significant in spectral-editing sequences, because spectral editing critically relies on narrow-band frequency selective pulses or accurate spectral alignments among scans for subtraction/addition of spectra. Frequency drift can occur due to subject’s movement and/or MR system instability. Even in advanced MR systems with self-shielded gradients, significant frequency drifts occur due to eddy current-induced heating and cooling of passive shim materials, particularly after MR scans with heavy gradient duty cycles. The effects of frequency drifts can be mitigated through prospective and retrospective frequency corrections. Currently, most spectral-editing methods use post-processing approaches to correct the effects of frequency drifts retrospectively. In this study, we have developed a prospective frequency correction method and implemented it in a semi-LASER based TE-averaged sequence for glutamate detection.

INTRODUCTION

Frequency drifts during MRS acquisition results in broad and distorted spectral lineshapes, a reduced SNR, and quantification errors. The consequence of frequency drifts is particularly significant in spectral-editing sequences, because spectral editing critically relies on narrow-band frequency selective pulses or accurate spectral alignments among scans for subtraction/addition of spectra. Frequency drift can occur due to subject’s movement and/or MR system instability. Even in advanced MR systems with self-shielded gradients 1, significant frequency drifts occur due to eddy current-induced heating and cooling of passive shim materials, particularly after MR scans with heavy gradient duty cycles. The effects of frequency drifts can be mitigated through prospective and retrospective frequency corrections 2-7. Currently, most spectral-editing methods use post-processing approaches to correct the effects of frequency drifts retrospectively 8-11. In this study, we have developed a prospective frequency correction method and implemented it in a semi-LASER based TE-averaged sequence 12,13 for glutamate detection.

METHODS

The prospective frequency correction was implemented by integrating an interleaved reference scan (IRS) method 2 into a semi-LASER based TE-averaged single-voxel sequence on a Skyra 3 T scanner (Siemens, Erlangen, Germany). The IRS method includes a navigator scan based on PRESS without water suppression and a low flip angle (10°) of the excitation RF pulse 2. Because our primary goal was to correct the frequency drifts, the spoiler gradients for water suppression were excluded from the navigator scan, allowing longer recovery time and reduced saturation effects. The navigator data containing the water signals were delivered to the on-line reconstruction, and the peak position of water signals was used to track the frequency drift in each scan. The frequencies of RF pulses and the receiver were updated using the measured frequency drift in navigator data prior to the next acquisition 5. The prospective frequency correction was tested on phantoms and human subjects after the realistic gradient heating by 30-min fMRI experiments with heavy duty gradient cycles. Parameters of the TE-averaged semi-LASER sequence with and without prospective frequency correction were: voxel size=3x3x3 cm3, TR=2250 ms, TE=35-355 ms in increments of 10 ms, NT=4, and bandwidth=1000 Hz.

RESULTS

The frequency drift of the water signal in a phantom was 3.2 Hz/min determined by the linear fit (Fig. 1a), and was effectively corrected with the prospective frequency correction (Fig. 1b). Equivalent frequency drifts of the phantom, 3.2 Hz/min, were measured from a healthy subject (Fig. 2a), although the frequency fluctuation was greater. The average fluctuations were ±0.29 Hz and ±0.11 Hz in the human and phantom scans, respectively. The greater fluctuations in the human scans are attributed to broader linewidths due to physiological motions and uncertainties in determining the water peak position. The effective corrections of frequency drifts in the human scans are shown in Fig. 2b. Distorted lineshapes of NAA at 2.0 ppm, creatine at 3 and 3.9 ppm in the uncorrected spectrum (Fig. 2b, top) have been clearly corrected after the prospective frequency corrections (Fig. 2b, bottom). Significantly improved linewidths and SNR were visible in the spectrum with the frequency correction, demonstrated by the narrower linewidth of NAA (5 Hz (corrected) vs. 13 Hz (uncorrected)). Better defined glutamate signals were also visible in the spectrum with the frequency correction.

DISCUSSION

The proposed prospective frequency correction method reliably corrected over 3 Hz/min frequency drifts occurred during MRS scans in both phantom and human scans. The effect of frequency drifts tend to be pronounced when the measurement requires longer scan times, as for metabolites with low concentration, e.g., GABA, GSH and vitamin C, using editing sequences. The prospective frequency correction method provides advantages over retrospective correction methods as this method can be applied to CSI data acquisition, which does not provide any spectral peak to track frequency drifts. This method also allows accurate spatial localization and consistent water suppression through real-time frequency updates of RF pulses. Combination with prospective motion correction should further improve the accuracy of localization and spectral quality during the presence of subject motion 4,14.

CONCLUSION

We demonstrated the enhanced spectral quality by the effective correction of gradient heating induced frequency drifts in a 1H MRS editing method using a prospective frequency correction method.

Acknowledgements

This work is partly supported by the National Institutes of Health (S10RR29577, UL1TR000001) and the Hoglund Family Foundation.

References

[1]. Foerster BU, et al, MRM (54), 1261-7, 2005.

[2]. Thiel T, et al, MRM (47), 1077-82, 2002.

[3]. Helms G, et al, MRM (46), 395-400, 2001.

[4]. Zaitsev M, et al, NMR Biomed (23), 325-32, 2010.

[5]. Lange T, et al, JMRI (33), 748-54, 2011.

[6]. Keating B, et al, MRM (68), 1339-45, 2012.

[7]. Near J, et al, MRM (73), 44-50, 2015.

[8]. Waddell KW, et al, MRI (25), 1032-8, 2007.

[9]. Evans CJ, et al, JMRI (38), 970-5, 2013.

[10]. Harris AD, et al, MRM (72), 941-8, 2014.

[11]. Bhattacharyya PK, et al, MRM (58), 808-12, 2007.

[12]. Hurd R, et al, MRM (51), 435-40, 2004.

[13]. Lin M, et al, MRM (71), 911-20, 2014.

[14]. Hess AT, et al, MRM (66), 314-23, 2011.

Figures

Fig. 1 a: Measured frequency drift in a water phantom over time. b: Final averaged spectrum with the prospective frequency correction.

Fig. 2 a: Measured frequency drift over time in a healthy volunteer. b: Spectrums obtained from the semi-LASER based TE-averaged scans without and with the prospective frequency correction.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
2371