4440
Direct comparison of fat navigators and Moiré phase tracking for retrospective brain motion correction at 7T1LIFMET, EPFL, Lausanne, Switzerland, 2Department of Biomedical Magnetic Resonance, Institute of Experimental Physics, Otto-von-Guericke-University, Magdeburg, Germany, 3CUBRIC, School of Engineering, Cardiff University, United Kingdom, 4German Center for Neurodegenerative Disease, Magdeburg, Germany, 5Center for Behavioral Brain Sciences, Magdeburg, Germany, 6Leibniz Institute for Neurobiology, Magdeburg, Germany
Synopsis
Retrospective rigid body motion correction based on FatNavs or MPT motion information are directly compared. Both modalities significantly improve image quality of very high resolution anatomical images, but both suffer from drawbacks: rigid marker fixation during long scans for MPT and low temporal resolution for FatNavs. Quantitative analysis confirms these visual observations.
Introduction
Subject motion remains a source of significant artefacts in MRI1. A variety of motion correction techniques has been presented to overcome this challenge. Using parallel imaging, fast fat-selective navigators (FatNavs) can quantify and retrospectively correct head motion2 without any additional hardware. Alternatively, Moiré Phase Tracking (MPT) can provide motion estimates by optically tracking a marker attached rigidly to the subject’s upper jaw3. Even though MPT requires additional hardware, it is commonly considered as the current gold standard tracking modality and enables prospective as well as retrospective correction.
In this work, we simultaneously tracked subject motion with MPT and FatNavs to compare their motion estimates as well as the quality of retrospective motion correction of high-resolution structural data.
Methods
The hardware used for the experiment included a 7T MR scanner (Siemens Healthcare, Germany), a 32-channel RF coil (Nova Medical Inc., USA) and the MPT setup (Metria Innovation, USA). Nine volunteers were scanned, after giving written consent, following the local ethics guidelines. They were asked to remain still during the scans. A custom-made mouth-piece carried the marker for optical motion tracking. MPT pose data were acquired with 86 fps.
Two isotropic 0.5 mm MP2RAGE4 volumes were acquired per subject (TI1/TI2/TR 800ms/2700ms/6s, FA1/FA2 7°/5° , R=2 in anterior-posterior direction and ¾ partial Fourier in left-right direction. TA=23 minutes 34 seconds). The FatNavs were acquired just after the second readout train (i.e. every plane of k-space). FatNavs parameters were: 2mm isotropic resolution, 3° flip angle, TE/TR=1.5ms/3.5ms, R=4x4 and ¾ partial Fourier in left-right and anterior-posterior directions. The calibration signal for GRAPPA reconstruction of the FatNavs was acquired at the very beginning of each scan.
Raw (no motion correction), FatNavs corrected and MPT corrected images were reconstructed offline and visually compared. Motion correction was implemented as multiplication by a phase factor (for translations) and a nuFFT5 operator (for rotations). For quantitative comparison, gradient entropy6 was computed for each inversion image. Relative changes in gradient entropy were reported w.r.t. uncorrected raw images.
Results and discussion
Figure 1 shows two examples of raw and FatNavs corrected images. For simplicity we only show the MP2RAGE “uniform” contrast and not the separate inversion images. Due to the high-resolution of the acquired images, the reader is encouraged to zoom in. With motion correction, blurring is reduced considerably and structures are visibly sharper. In this case, as in most of the scans, the MPT corrected and FatNavs corrected images showed no clear differences.
However, exceptions to this observation were found.
Figure 2 (Volunteer 2 scan 1) shows a case where the FatNavs
corrected data were sharper than the MPT corrected data,
as can be seen in the cerebellum and looking at the vessels delineation. On the
other hand, the FatNavs implementation in the MP2RAGE sequence has a
much lower temporal resolution than MPT, and can therefore not
account for fast
breathing, as is demonstrated in Figure 3 (Volunteer 8). In this
case, ringing is effectively suppressed in the MPT correction unlike
in the FatNavs correction.
The rigid body motion parameters of the two detection methods are shown in Figure 4. Overall patterns coincide between both modalities, but sometimes to different degrees. This might be due to the difficulty of keeping the marker truly co-moving with the head for long duration scans, or misalignments of the navigators, due to non-rigid motion of soft skin relative to the brain. For example, at the end of the second scan for volunteer 1, it is very plausible that the marker moved independently from the head, as the FatNavs showed the skull remained very static during that time. The case of volunteer 2 (both scans) potentially illustrates marker fixation as the main drawback of MPT. Figure 5 shows the relative gradient entropy change compared to the raw reconstruction, in percent. These values fit well with previously mentioned visual observations (see volunteers 2 and 8 for example).
Finally, it is plausible that prospective MPT based-correction can further reduce the artefacts level, maybe even more so for accelerated acquisitions. However, direct comparison with retrospective FatNavs-based correction would then not be possible.
Conclusion
We have directly shown that both FatNavs and MPT significantly improved image quality after retrospective motion correction. MPT remains vastly superior in terms of temporal resolution, but can possibly suffer from independent marker motion.
For the studied cohort (i.e. experienced and compliant subjects), both methods generally lead to equivalent retrospectively motion corrected images.
Acknowledgements
This work was supported by the NIH, grant number 1R01-DA021146, and in part by the Centre d’Imagerie BioMedicale (CIBM) the UNIL, UNIGE, HUG, CHUV, EPFL, the Leenaards and Jeantet Foundations, as well as the Swiss National Foundation through Grant 205321_153564.References
1. Zaitsev M, Maclaren J, Herbst M. Motion artifacts in MRI: A complex problem with many partial solutions. Journal of Magnetic Resonance Imaging. 2015;42(4):887–901.
2. Gallichan D, Marques JP, Gruetter R. Retrospective correction of involuntary microscopic head movement using highly accelerated fat image navigators (3D FatNavs) at 7T. Magnetic Resonance in Medicine. 2015;00:1–10.
3. Stucht D, Danishad KA, Schulze P, Godenschweger F, Zaitsev M, Speck O. Highest Resolution In Vivo Human Brain MRI Using Prospective Motion Correction. PloS one. 2015;10(7):e0133921.
4. Marques JP, Kober T, Krueger G, van der Zwaag W, Van de Moortele P-F, Gruetter R. MP2RAGE, a self bias-field corrected sequence for improved segmentation and T1-mapping at high field. NeuroImage. 2010;49(2):1271–1281.
5. J. A. Fessler and B. P. Sutton, “Nonuniform fast fourier transforms using min-max interpolation,” IEEE Trans. Signal Process., vol. 51, no. 2, pp. 560–574, Feb. 2003.
6. McGee KP, Manduca A, Felmlee JP, Riederer SJ, Ehman RL. Image metric-based correction (Autocorrection) of motion effects: Analysis of image metrics. Journal of Magnetic Resonance Imaging. 2000;11(2):174–181.
Figures
