Physicists and physicians interested in magnetic resonance elastography (MRE).MRE is capable of generating image contrast based on the viscoelastic properties of tissue by inducing and detecting time harmonic shear waves in the body [1]. Recent work in MRE of the brain suggested the exploitation of low vibration frequencies since shear waves in the regime of 20 Hz are nearly unaffected by attenuation [2]. Furthermore, it was stated that low drive frequencies are less onerous to patients and volunteers than high frequencies, in particular when combined with remote wave excitation as proposed in [3]. Based on these reports, we introduce a low-frequency protocol for brain-MRE including a remote compressed-air actuator which is placed beneath the shoulders and we test the reproducibility of this protocol.To introduce low-frequency MRE of the brain based on remote excitation of intracranial shear waves by a pressurized-air actuator and to analyze reproducibility of this method as well as inter-subject variation and day-to-day variation of the brain shear modulus in the low frequency regime.

An MRE setup was developed utilizing a single shot spin-echo EPI sequence based on retrospective wave phase correction as detailed in [2], which ensures an uninterrupted shear wave propagation through the tissue with minimized image acquisition time. Intracranial shear waves were excited by a remote actuator placed under the shoulders of the volunteers and operated by 17.85, 20.00 and 22.73 Hz drive frequencies (56, 50 and 54 ms cycle time). The rubber membrane of the actuator (11 cm diameter) was vibrated by air pressure pulses supplied with compressed air of approximately 0.25 bar from a medical air pipeline and controlled by a high-speed electromagnetic valve [4]. Imaging parameters: 1.5T Siemens Sonata MR scanner, 15 consecutive coronal slices, 2 mm isotropic voxel size, 8 wave dynamics, 3 wave field components, rectangular 1^st momentum nulled motion encoding gradient with 25 mT/m amplitude (38.4 ms duration), TR/TE=2500/92 ms, FoV=192×160 mm², matrix size: 96×80, total measurement time including three frequencies: 3:06 min

To analyze the reproducibility of the method, six healthy volunteers were scanned on three days within one week. On each of the three days, nine MRE-repetitions were acquired in three sessions (in total 27 repetitions per subject). After the third and the sixth repetition of each day, the subject left and reentered the scanner, and actuator and subject were repositioned.

In addition to the MRE scans, a magnetization-prepared rapid gradient echo (MPRAGE) sequence (TR/TE=2110/4.4 ms, TI=1100 ms, flip angle 15°, 1 mm isotropic voxel size) was acquired before each session.

MRE data were analyzed by multifrequency dual elasto-visco (MDEV) inversion as described in [2] yielding a 3D parameter map corresponding to the magnitude |G*| of the complex shear modulus.

Despite long scan times due to multiple repetitions, no discomfort was reported by any subject. As visualized by the three curl-components of the wave field at 20 Hz in Fig. 2 (bottom), high shear wave amplitudes were achieved throughout the whole brain. The |G*|-maps agree well to the anatomy. Figure 3 shows |G*|-values average within the cerebrum for the 27 scans of each subject. The variances along the repetitions are displayed in Table 1. The variance within one session (group averaged coefficient of variation CV_median = 0.8%) was smaller than the variance within one day (group averaged CV_median = 1.8%) and smaller than the total variance accounting for 27 measurements (2.5%). We note inter-subject differences raising the question to what extent |G*| reflects intrinsic structural differences between individual brains.

Low frequency MRE vibrations were felt less intense than intrinsic scanner vibrations. Furthermore, the remote actuator does not require any space within the narrow geometry of head coils resulting in a flexible setup. Our method shows good test-retest reproducibility. Higher variances between days as compared to intra-day scans indicate the influence of day-to-day changes of brain viscoelasticity. Since low-frequency MRE is potentially more sensitive to fluid-solid interactions in poroelastic media, vascular effects may play a role here. This may contribute to the observed inter-subject differences and should be further investigated. Another source of variability is the delineation of regions which was done with the help of atlas-based image registration in our study. This approach neglects boundary effects near tissue interfaces [5].

In summary, cerebral low frequency MRE provides a novel source of mechanical information of brain tissue which is attainable in a fast and reproducible way with less onerous head stimulations as required by conventional MRE. The measured elastograms reveal individual and day-to-day variations of brain viscoelasticity requiring further investigations.