Synopsis
During macromolecule (MM) suppressed GABA MRS acquisition, subject motion may cause the spectra to be acquired at an incorrect region of interest and with suboptimal shim. Furthermore, effective MM-suppression requires the editing pulses to be applied consistently at 1.7 ppm, necessitating real-time frequency updates, which can be exacerbated in the presence of motion. We demonstrate that a pair of 3D EPI volumetric navigators acquired once per TR is able to perform accurate motion and magnetic field inhomogeneity correction in real time during MM-suppressed MEGA-SPECIAL GABA MRS.Purpose
MEGA-SPECIAL
(MSpc) is a magnetic resonance spectroscopy (MRS) sequence comprising of longer
frequency selective editing pulses, allowing γ-Aminobutyric acid (GABA) acquisition
without macromolecules (MM) at TE of 68 ms
(1,2). Unlike the
ubiquitous MEGA-PRESS sequence
(3), MSpc requires four acquisitions
per localized edited spectrum, rendering it more sensitive to subject motion
and magnetic field (B
0) inhomogeneity. Moreover, the MM-suppression
technique is highly dependent on B
0 field stability, necessitating
real-time frequency update to ensure frequency selective pulses are applied
consistently at the resonance of 1.7 ppm
(1). The aim of this work is
to measure and correct simultaneously in real-time for head position and
magnetic field inhomogeneity in terms of zero and first order shim gradients
using volumetric navigators (vNav)
(4).
Methods
A pair of 3D multishot EPI vNav
was added to the standard MSpc sequence (vNavMSpc).
Motion parameters were estimated from the
magnitude images of vNav, while shim parameters were measured from EPI field
map, which was reconstructed from magnitude and phase images of vNav. The MSpc
sequence receives in real time these parameters to update for subject motion
and B
0 distortion to all pulses including the localization,
MEGA-editing, OVS and VAPOR pulses
(1).
All scans were performed on a Siemens Allegra 3T scanner. Three subjects were
scanned with the vNavMSpc sequence as follows: i) Reference (no intentional
motion) with no correction (NoCo), ii) Reference with shim and motion
correction (ShMoCo), iii) intentional motion with ShMoCo, and iv) intentional
motion with NoCo. The vNav protocol was
as follows: 3D encoded EPI, 32 x 32 x 28, flip angle 2°, TR 16 ms, TE 6.6/9 ms.
The vNavMSpc parameters were: (3 cm)
3 voxel positioned in a
mid-parietal, 2048 points, bandwidth 2kHz, 148 averages and TR/TE 4000/68 ms.
GABA was acquired without MM contamination.
The duration of every scan was 10 minutes 8 seconds. The motion involved
chin up-down and chin left-right rotations of about 5°, which also resulted in
translations of about 4 mm (Figure 1). After every data acquisition involving
motion, subjects were instructed to return to their original positions. All data were processed using the Gannet
toolkit
(5). Residual fitting error (FitErr) and GABA/Cr
concentration – measured by Gannet – were compared between data acquired in the
absence and presence of intentional motion.
Results
Figure
2 shows spectra from the four different acquisitions of one subject. The edited
spectra from the NoCo and ShMoCo reference scans have comparable and excellent
spectral quality. Motion caused extreme frequency drifts, which led to
overestimated GABA amplitude (arrow A in Fig 2) and insufficient suppression of
the NAA peak (arrow B in Fig 2) of the edit-on spectrum. The NoCo motion scans
show increases in all parameters (FitErr: 15.22 ± 2.43; GABA/Cr: 0.22 ± 0.17),
while ShMoCo recovers the data yielding estimates similar to the reference scans
(FitErr: 9.98 ± 1.92; GABA/Cr: 0.037 ± 0.01).
Discussion
Head pose and shim estimates were estimated in
each TR using the vNav and if any motion was detected, the correction was
applied in the following TR. The NoCo scans during motion led to lower spectral
quality and overestimated GABA/Cr concentrations. The ShMoCo scans effectively
reduced the artefacts and acquired well edited GABA peaks. The ubiquitous
technique for correction is retrospective frequency and phase correction
(6).
Although this method reduces subtraction artefacts, localization errors remain
uncorrected and corrupted signal arising from poor shimming cannot be corrected
retrospectively. Keating et al.
(7) implemented PROMO and a navigator
as an efficient alternative method to simultaneously correct in real-time for
subject motion and B
0 inhomogeneity. However, this technique
requires cameras and a mouthpiece affixed to the subject, necessitating subject
compliance. Real-time motion and shim correction have been applied in MRSI
MEGA-LASER sequence for GABA+MM acquisition using volumetric navigators
(8).
Our implementation involves MM-suppression technique, demanding stronger B
0
stability, which can be impaired by the presence of motion especially when
using motion sensitive MSpc sequence. There are no direct comparisons between
MEGA-LASER or MEGA-sLASER and MSpc sequences, but when referenced to MEGA-PRESS
(3),
MEGA-LASER and MEGA-sLASER demonstrate lower spatial variations, are less
sensitive to motion and yield more GABA signal than MSpc
(1,8-9).
This suggests that the real-time motion and shim correction is a necessity for
MSpc sequence to ensure proper C4-GABA acquisition.
Conclusion
The
vNav performed accurate motion and shim correction in real-time without increasing
the scan time and resulted in well edited GABA spectra. This technique can
greatly benefit GABA MRS, which is challenging due to low signal often necessitating
long acquisition times.
Acknowledgements
The
South African Research Chairs Initiative of the Department of Science and
Technology and National Research Foundation of South Africa, Medical Research
Council of South Africa, NIH grants R21AA017410, R21MH096559, R01HD071664.References
(1)
Near J, Simpson R, Cowen P, et al. Efficient γ-aminobutyric acid editing at 3T
without macromolecule contamination: MEGA-SPECIAL. NMR in Biomed. 2011,
24(10):1277-1285. (2) Henry PG, Dautry C, Hantraye P, et al. Brain GABA editing without macromolecule
contamination. Magn Reson Med. 2001, 45(3):517-520. (3) Mescher M, Tannus A,
Johnson M et al. Solvent suppression using selective echo dephasing. JMR Series
A 1996, 123(2):226-229. (4) Hess AT, Dylan Tisdall M, Andronesi OC, et al.
Real-time motion and B0 corrected single voxel spectroscopy using volumetric
navigators. Magnetic resonance in medicine 2011, 66(2):314-323. (5) Edden RA, Puts NA,
Harris AD, et al. Gannet: A batch-processing tool for the quantitative analysis
of gamma-aminobutyric acid–edited MR spectroscopy spectra. JMRI. 2013. (6) Near
J, Edden R, Evans CJ, et al. Frequency and phase drift correction of magnetic
resonance spectroscopy data by spectral registration in the time domain. MRM.
2015, 73:44-50. (7) Keating B, Ernst T: Real-time dynamic frequency and shim
correction for single-voxel magnetic resonance spectroscopy. MRM. 2012,
68:1339-1345. (8) Bogner W, Gagoski B, Hess AT, et al. 3D GABA imaging with
real-time motion correction, shim update and reacquisition of adiabatic spiral
MRSI. Neuroimage 2014, 103:290-302. (9) Andreychenko A, Boer VO, Arteaga de
Castro CS, et al. Efficient spectral editing at 7 T: GABA detection with MEGA-sLASER.
MRM. 2012, 68:1018-1025.