Rotation optimization for semi-LASER implementation under the constraint of maximum gradient strength
Ralph Noeske1

1GE Healthcare, Potsdam, Germany

Synopsis

The semi-LASER sequence is less prone to chemical shift displacement errors and shows higher B1-robustness compared to PRESS. To achieve short echo times high amplitude crusher gradients are used to suppress unwanted coherence signals. Goal of this work was to implement an algorithm that prevents violation of gradient strength constraints due to unrestricted rotation of the voxel while optimizing for shortest possible echo time.

PURPOSE

Due to the low bandwidth of the refocusing pulses of the Point-Resolved Spectroscopy (PRESS) sequence at short echo times there are larger chemical shift displacement errors (CSDE) at high B0 field (≥ 3T). The use of pairs of adiabatic selective refocusing pulses within the semi-LASER sequence1-3 reduces the CSDE and increases the B1-robustness. To achieve a short echo time ≤ 30ms short RF pulses with high B1 amplitude as well as short crusher gradients with high amplitude have to be used. A previously reported crusher scheme1 reduces unwanted coherence signals with an optimized set of crusher gradients. Rotation of the prescribed voxel can lead to higher gradient amplitudes on the physical axis that can result into values above the maximum gradient strength depending on system constraints. A common way to prevent this is to set the maximum gradient strength on each logical axis for the worst case of a rotation of 45° in each direction leading to a reduction of the maximum allowed gradient strength by a factor of √3 and therefore to longer crusher gradients if the area under the crusher gradients should be kept constant. The goal of this work was to implement an algorithm that keeps the areas of the crusher gradients constant and takes the rotation of the voxel into account to give shortest possible TE for any prescription under the constraint of maximum gradient strength on each physical axis.

METHODS

The semi-LASER sequence was implemented following the crusher scheme described in1 (Fig. 1). An asymmetric slice selective excitation pulse (bandwidth 3.7kHz, 4.2ms pulse duration) and adiabatic refocusing pulses with a trapezoid amplitude envelope2 (bandwidth 8kHz, 4.5ms pulse duration) were used. The implemented optimization algorithm kept the areas of the crusher gradients constant. A rotation of the voxel is defined by the rotation matrix that transform the gradient amplitudes from the logical (read, phase, slice) into the physical coordinate system (x,y,z). With this rotation matrix the gradient amplitudes in the physical coordinate system were calculated and checked against the constraint of maximum gradient strength in each of the 5 sections (Fig. 1). In case of a violation, first the crusher gradient that was shortest in this section got stretched, if this still led to a violation all 3 crusher gradients in that section got stretched. Phantom (GE MRS sphere) and in-vivo experiments were performed on a 3.0T MR750w system (GE Healthcare) with a maximum gradient strength of 33mT/m and a slew rate of 120T/m/s. Acquisition protocol: 2x2x2cm3 voxel size, 32-step phase cycling, 64 averages.

RESULTS

Table 1 shows crusher amplitudes in logical and physical coordinate system as well as the inter-pulse times for various prescriptions. The minimum TE of 29ms extends slightly with increasing angulation of the voxel in multiple directions. Fig. 2 shows the corresponding phantom spectra and Fig. 3 in-vivo spectra of a left anterior white matter voxel.

DISCUSSION

In the proposed crusher scheme of the semi-LASER implementation especially the crusher gradients after the last adiabatic refocusing pulse have high amplitudes on all 3 axis. Even under the constraint of a maximum gradient strength that is close to this amplitude the implemented algorithm just slightly increases the minimum echo time from 29ms to 30ms for the worst case of rotation of the voxel by 45° in all 3 directions but keeping it at the minimum echo time for most common angulations and 10% shorter compared to the common way of reducing maximum gradient strength by a factor of √3 (29ms vs. 32ms).

Acknowledgements

The author thanks Uzay Emir, Gülin Öz and Tom Scheenen for providing information of the semi-LASER crusher scheme.

References

1. Öz G et al. Short-Echo, Single-Shot, Full-Intensity Proton Magnetic Resonance Spectroscopy for Neurochemical Profiling at 4 T: Validation in the Cerebellum and Brainstem. Magn Reson Med. 2011; 65: 901-10

2. Boer et al. 7-T 1H MRS with adiabatic refocusing at short TE using radiofrequency focusing with a dual-channel volume transmit coil. NMR Biomed. 2011; 24(9): 1038-46

3. Scheenen TW et al. Short Echo Time 1H-MRSI of the Human Brain at 3T With Minimal Chemical Shift Displacement Errors Using Adiabatic Refocusing Pulses. Magn Reson Med. 2008; 59:1-6

Figures

Fig. 1: Implemented semi-LASER sequence with the 5 sections (roman letters)

Table 1: Crusher amplitudes in logical and physical coordinate system of the last crusher gradient (5th section) and inter-pulse times for various prescriptions. TE increases only in the worst case of triple angulation of 45° each

Fig. 2: Phantom spectra acquired with the semi-LASER sequence and different voxel angulations. (a) axial, (b) oblique-axial, (c) double-oblique and (d) triple-oblique (45° each)

Fig. 3: In-vivo spectra of a left anterior white matter voxel. (a) axial, (b) oblique-axial and (c) double oblique



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
4002