The periodic volume changes accompanying the arterial blood entering the brain cause a periodic brain motion, which in turn creates a pulsatile component in the cerebrospinal fluid (CSF) flow^1-4. As the mechanical link between blood and CSF, the degree of brain motion may serve as a biomarker for some neuropathologies. In addition, the ability to accurately characterize brain motion helps increase our understanding of the magnitude of these motion effects, and improve imaging techniques and therapies sensitive to motion such as diffusion imaging⁵ and stereotactic radiosurgery (SRS) treatment⁶.

Displacement encoding with stimulated echoes (DENSE) is a motion imaging technique that encodes pixel-wise tissue displacements into the phase of the stimulated echoes^7,8. DENSE is well suited for measuring small displacements, and has been applied to measure brain motion as low as 0.01 mm^9,10. Previous studies showed different motion magnitude of brain regions^9,10, but to our knowledge, no motion range grouping has been attempted. In this study, we took advantage of the high motion sensitivity of DENSE to quantify the displacements of different brain regions, and hypothesized that the brain regions near the CSF had larger motion magnitude than regions far away.

Approved by our IRB, after informed consent was obtained, 9 healthy volunteers (33.7 ± 11.0 years, two females) were scanned on a 3T scanner with the head and neck coils (Tim Trio, Siemens, Erlangen, Germany). Each subject was in the supine position with simple immobilization buffers around the head. A mid-sagittal slice through the cervical cord and brain stem was imaged with a customized, pulse-gated, segmented EPI, cine DENSE sequence. Image parameters included displacement encoding frequency k_e = 1.5 cycle/mm, through-plane dephasing frequency k_d = 0.08 cycle/mm, TE = 8.9-10.4 ms, TR = 55-59 ms, EPI factor = 8, segments = 16, pixel size = 1.2 × 1.2 mm², slice thickness = 7 mm, averages = 4, frames = 13-16, where parameter variations depended on the pulse durations. The subjects were instructed to remain motionless during the scan.

The DENSE images were reconstructed inline to generate DICOM images with displacement encoded in two orthogonal directions¹¹, and then these images were exported for offline processing using ImageJ (National Institute of Health, Bethesda, MD, USA). Briefly, the phase-reconstructed image in each direction was divided by 2πke to convert to displacement, and then the 2D displacement map was obtained by combining data in two directions. Regions of interest (ROIs) were placed in frontal lobe, occipital lobe, parietal lobe, cerebellum, midbrain, pons, medulla and optic chiasm, respectively. The mean 2D displacement values in the ROIs for each frame were recorded for all 9 volunteers, and fitted using a 3rd or 4th order polynomial in Matlab (The Mathworks, Natick, MA, USA) to compensate for varying temporal resolutions in the acquired data. An analysis of variance (ANOVA) was performed using R v3.2 (R Core Team, Vienna, Austria) to test and group displacement as a function of brain regions. Post-hoc comparisons were performed using Tukey’s HSD to adjust for multiple comparisons.

Example DENSE magnitude-reconstructed image with the ROIs and their grouping letters (explained later) are shown in Fig 1. Example 2D displacement maps at several time points after the pulse triggering are shown in Fig. 2. Fig. 3 shows the fitted displacement-time curves from 9 volunteers. And Fig. 4 summarizes and groups the peak displacement averaged across the 9 volunteers. ANOVA found that the model was highly significant (p < 0.0001). The post-hoc comparisons identified 3 overlapping groups of locations with respect to peak displacement with group “a” having the least displacement and group “c” having the most. In general, the brain regions near the CSF (midbrain, pons, medulla and optic chiasm) had larger motion magnitude than regions far away (frontal lobe, occipital lobe, parietal lobe and cerebellum). These were consistent with our observations in both Fig. 2 and Fig. 3.In this work, we have demonstrated that the DENSE technique can be used to measure and group motion of different brain regions. Preliminary results in 9 volunteers showed that the brain regions near the CSF (midbrain, pons, medulla and optic chiasm) had larger motion magnitude than regions far away (frontal lobe, occipital lobe, parietal lobe and cerebellum). DENSE enables us to investigate brain motion to a level of detail that has not been previously possible. The findings in this study may bring new insight into brain motion and provide useful information to improve potential imaging and therapy techniques.