­­­Fast and flexible 3D-EPI fat navigators for high-resolution brain imaging at 7 Tesla
Pieter F Buur1, Wietske van der Zwaag1, José P Marques2, and Daniel Gallichan3

1Spinoza Centre for Neuroimaging, Amsterdam, Netherlands, 2Donders Institute for Brain, Cognition and Behaviour, Radboud University Nijmegen, Nijmegen, Netherlands, 3Centre d'Imagerie BioMédicale (CIBM), EPFL Lausanne, Lausanne, Switzerland


Motion correction using interleaved fat navigators is a promising approach for high-resolution brain imaging at 7 Tesla. We have implemented a 3D-EPI fat navigator to reduce acquisition time and thereby minimize overhead in sequences with little or no dead time. The efficacy of motion-induced artefact removal using the fat navigators is demonstrated for 0.6 mm isotropic inversion-prepared (MPRAGE) and 0.5 mm isotropic non-prepared 3D TFE (GRE) protocols.


To minimize the acquisition time of fat-only image navigators in order to allow motion correction of high-resolution anatomical images with reduced scanning overhead.

The gain in signal-to-noise ratio at ultra-high field MRI enables very high resolution structural and functional imaging of the human brain. At these resolutions (~0.5mm and smaller), even minimal head movement, which is unavoidable especially in older and clinical populations, can cause significant artefacts and blurring that obscure the fine detail in the images. Recently, a navigator-based approach employing fat-selective excitation was introduced1 that allows for estimation and retrospective correction of submillimeter head motion. One drawback of the method is the relatively long acquisition time of the navigator. The current implementation builds on the previous method by using an undersampled 3D-EPI readout similar to2, to drastically shorten the acquisition time of the navigator, thereby allowing for a more flexible use of the fat navigators and reduced overhead in sequences without any (or with reduced) dead time such as gradient echo (GRE) sequences.


Data were collected on a 7T MRI scanner (Philips Healthcare, Cleveland, Ohio, USA) with a 32-channel head coil (Nova Medical, Wilmington, Massachusetts, USA). Fat navigators with a 3D-EPI readout were implemented as follows: fat-selective excitation using a 121-binomial pulse scheme, FA 1°, TR/TE 17/5.6 ms, 2 mm isotropic voxels, sagittal slice orientation, FOV 240x240x160 mm (FHxAPxRL), SENSE factor 4x2 (APxRL) and partial Fourier factor 0.75 in both PE directions, resulting in a volume TR of 450 ms. Navigators were inserted into (1) a 0.6 mm isotropic inversion-prepared 3D TFE (MPRAGE) and (2) a 0.5 mm isotropic (non-prepared, GRE) TFE sequence using the Multiple Instantaneous Switchable Scans (MISS) functionality available on the Philips platform. Imaging parameters for (1): TR/TE/TI 6.2/2.3/1300 ms, FA 7°, FOV 195x224x156 mm, SENSE 2x2, 128 TFE shots, TFE interval 4500 ms, including the navigator (Figure 1). Imaging parameters for (2): TR/TE 19/15 ms, flip angle 7°, FOV 195x224x156 mm, SENSE 2x2, 156 TFE shots, TFE interval 4500 ms, including the navigator. In both sequences, navigators were acquired every TFE shot (i.e. k-plane) resulting in a total acquisition time of 10 and 11 minutes, respectively. For the non-prepared TFE, the shot length was chosen such that the scanning overhead due to the navigator acquisition was only 10%. Three healthy volunteers were scanned, and instructed to gradually move their head over the course of the acquisition. Motion parameters were estimated from the fat navigator images using SPM8 and subsequently used to correct the raw k-space data as previously described1. Data shown has been bias-field corrected.


Figure 1 shows the pulse sequence diagram for the interleaved sequence as well as an example of a 3D-EPI fat navigator image. Note that a fast navigator is especially important for the non-prepared TFE. Figure 2 shows the motion parameters and corresponding uncorrected and corrected T1-weighted images for three subjects. Figure 3 shows the motion parameters for the non-prepared acquisitions, as well as the corresponding image with and without motion correction for two subjects. Zoomed areas of the main panel show the correction in more detail. Although subjects were explicitly asked to move, displacements were similar to those seen in naive subjects (1-2 mm overall, 1-6 degrees rotation). The corrected images clearly show the beneficial effect of incorporating the fat navigator-estimated motion parameters into the image reconstruction for both sequences. Motion effects are especially prominent in the uncorrected images at boundaries, near vessels and at the cortical surface.


Fat navigators form a versatile, cost-effective and easily implemented alternative to more elaborate motion correction methods relying on external hardware3. The sequence interleaving capabilities of the MISS functionality very straightforwardly allow the acquisition of fat-images interspersed with any host sequence. Here, we have chosen to demonstrate the effectiveness of the 3D-EPI based fat images in the T1w-TFE/MPRAGE and TFE/GRE sequences. Incorporation in other 3D sequences such as MP2RAGE or TSE sequences would also be possible. Judging from the increased image quality of the motion corrected data, 3D-EPI based fat navigators clearly show potential for reliably detecting motion. The shorter readout of the 3D-EPI allows the acquisition of a fat-navigator image in 450ms, compared to the 1.2s used previously for a GRE-readout at the same nominal resolution1, leading to acceptable overhead in continuous acquisitions such as the TFE and TSE.


We successfully implemented a fat navigator scheme on a Philips platform, whereby the use of a 3D-EPI readout allows shorter fat navigator acquisition times. Successful correction of 1-5 mm motion was achieved in two different image types, namely T1w-TFE and non-prepared TFE.


No acknowledgement found.


1. Gallichan, D., Marques, J. P. and Gruetter, R. (2015), Retrospective correction of involuntary microscopic head movement using highly accelerated fat image navigators (3D FatNavs) at 7T. Magn Reson Med. doi: 10.1002/mrm.25670

2. Mårtensson M, Engström M, Avventi E, et al. (2015). 3D FatNav: Prospective Motion Correction for Clinical Brain Imaging. ISMRM 2015, #0816.

3. Maclaren, J., Herbst, M., Speck, O. and Zaitsev, M. (2013), Prospective motion correction in brain imaging: A review. Magn Reson Med, 69: 621–636.


Figure 1. (a) Pulse sequence diagram for the interleaved TFE and 3D-EPI acquisition scheme. (b) Example (mean) fat navigator image.

Figure 2. Motion correction results for the T1w-TFE sequence in three subjects. (a) Fat navigator-based motion estimates. (b, c) Example slices for uncorrected and corrected data, respectively. (d, e) Details from the slices in (b, c).

Figure 3. Motion correction results for the non-prepared TFE sequence in three subjects. (a) Fat navigator-based motion estimates. (b, c) Example slices for uncorrected and corrected data, respectively. (d, e) Details from the slices in (b, c).

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