Synopsis

Keywords: Pulse Sequence Design, Non-Proton, High-Field MRI, Body, 23Na MRI, Field Mapping

²³Na MRI is a promising imaging method, but suffers from lower SNR, hence from longer acquisition times and lower image resolution than ¹H MRI. Especially in the torso, motion and magnetic field inhomogeneities impede quantitative analysis of ²³Na concentrations. To tackle this, a new pulse sequence is presented that yields self-gated respiratory-sorted B₁⁺ and B₀ maps from a single measurement. B₁⁺ mapping with the new sequence is less prone to motion artefacts, since k-space projections of two flip angles are acquired in an interleaved manner. This reduces the influence of varying respiration, which can lead to artefacts in conventional mapping.

Introduction

Sodium plays an important role in many physiological processes, motivating its quantitative measurement in vivo¹. However, ²³Na MRI in the torso can be especially challenging due to respiration and B₀ or B₁ inhomogeneities, for which quantitative ²³Na images should be corrected.

This study presents a pulse sequence that enables self-gated respiratory-sorted B₀ and B₁⁺ mapping from a single measurement based on the phase difference² and double angle methods (DAM)³. To reduce the influence of varying respiratory motion, the k-space projections of two flip angles are acquired in an interleaved manner. The obtained maps can be applied to correct for B₀ and B₁⁺ field inhomogeneities in quantitative ²³Na MRI. A comparison between B₀ and B₁⁺ maps acquired with the new sequence and conventional mapping is performed in a respiratory phantom and the human torso.

Methods

Alternating excitation with 45° and 90° pulses was implemented in a 3D density-adapted radial sequence⁴ with a double-echo readout and a golden angle distribution of projections⁵ (schematic sequence diagram in Fig. 1A). Further, each projection acquires two samples in the k-space center to extract the respiratory motion⁶ via an ADC shift of two dwell times.

All measurements were performed on a 7T whole-body MR scanner⁷ with a 4-channel oval-shaped birdcage ²³Na coil⁶.

To evaluate the general applicability of the new sequence, B₀ and B₁⁺ maps were obtained using the phase difference method and the DAM³ in a stationary respiratory phantom and compared to a reference (three measurements with: 1. 45°, TE1; 2. 90°, TE1; 3. 90°, TE2). The phantom (25mM NaCl) has a hole through its length, in which a smaller phantom (58mM NaCl)⁸ was inserted. The smaller phantom was placed in two different positions to simulate two discrete positions within a breathing cycle. In a second step, the small phantom was moved by hand in the rhythm of the mover’s own respiratory frequency to simulate respiratory motion to test the sequence under moving conditions. Additionally, an in vivo measurement was performed in a healthy volunteer (female, 32 years).

Parameters (phantom): TE1/TE2=1.65ms/7.6ms, TR=300ms, t_pulse=3ms, nominal resolution: (17mm)³, number of projections N=1150, acquisition time=11.5min, dwell time=13µs.

Parameters (in vivo): TE1/TE2=1.65ms/7.6ms, TR=300ms, t_pulse=3ms, nominal resolution: (17mm)³, number of projections N=1550, acquisition time=15.5min, dwell time=13µs.

Relative B₁⁺ maps were determined with the DAM³ from the data of the first echo after the 45° and the 90° excitation, respectively. With the data from both echos (TE1, TE2) after the 90° excitation, a B₀ map was calculated (Fig. 1B). For each measurement the respiratory signal was calculated by using the two datapoints at the k-space center of each projection⁶ (Fig. 2). All projections were sorted into two respiratory-sorted subsets depending on whether the respiratory signal of each projection was ≤0 or >0. These data subsets were reconstructed with a NUFFT⁹ and subsequently used to calculate the B₀ and B₁⁺ maps.

Results

²³Na MR images, B₀ maps and relative B₁⁺ maps of the stationary phantom are shown for the new mapping sequence in Fig. 3A+B with their corresponding reference. Maps for the new sequence and the reference are in good agreement in the stationary case (Fig. 3C).

In Fig. 4 B₀ and B₁⁺ maps are shown that were generated from the ²³Na images of the moving phantom measured with both the new sequence and reference. The unsorted maps of the new sequence and the reference show similar field distributions, while the sorted B₁⁺ maps of the reference show (contrary to the new sequence) a slight deviation from the classical birdcage B₁⁺ distribution (compare Fig. 3B). B₀ and B₁⁺ maps in breathing state 1 measured with the new sequence show better agreement with the stationary maps in state 1 (Fig. 3B/4) than the reference.

Comparing the relative B₁⁺ maps of the new mapping sequence and the reference in vivo (Fig. 5), all maps (sorted/unsorted) show similar distributions for the new sequence, whereas in the reference the sorted maps differ from the unsorted.

Discussion and Conclusion

Self-gated respiratory-sorted B₀ and B₁⁺ mapping from a single measurement was shown to be feasible. Respiratory-sorted relative B₁⁺ maps of the reference show a greater deviation from the classical birdcage pattern than those measured with the new sequence. Additionally, the respiratory-sorted B₀ and B₁⁺ maps of the new sequence in the moving respiratory phantom have a higher correspondence with the stationary images of the phantom, and the sorted in vivo B₁⁺ maps with the unsorted map. This is due to the advantageous timing of the new sequence: With the new sequence the same k-space projection needed for the B₀ and B₁⁺ maps is acquired for each projection in ≤300ms, whereas in the reference this time difference (≥5min 45s) is much higher (Fig. 1B). While the breathing state (exhaling/inhaling) is likely to be the same or at least similar within 300ms, it is presumable that the depth of inhalation and exhalation will change over 5min 45s. Therefore, the MR images of the different breathing states used to determine the field maps have a higher correspondence with each other when measured with the new sequence. Thus, the new mapping sequence reduces the influence of varying respiration that could lead to artefacts in conventional mapping.

Acknowledgements

No acknowledgement found.