5009

Acquisition protocol for glucoCESL MRI in the human brain at 7T with reduced motion-induced artifacts

Philip S. Boyd¹, Johannes Breitling¹, Ferdinand Zimmermann¹, Andreas Korzowski¹, Moritz Zaiss², Kai Herz², Patrick Schuenke¹, Nina Weinfurtner³, Heinz-Peter Schlemmer³, Mark E. Ladd¹, Peter Bachert¹, Daniel Paech³, and Steffen Goerke¹

¹Division of Medical Physics in Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany, ²High‐field Magnetic Resonance Center, Max Planck Institute for Biological Cybernetics, Tübingen, Germany, Tübingen, Germany, ³Department of Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany

Synopsis

In this study, an acquisition protocol for glucoCESL MRI examinations in the human brain is presented, that was optimized for suppression of motion-induced artifacts. This was achieved by using a combination of (i) a 3D imaging readout, which enabled co-registration of the acquired images over the course of time, and (ii) a ΔR_1ρ contrast based on quantitative R_1ρ maps instead of R_1ρ-weighted images. The presented acquisition protocol improves the applicability of glucose-weighted MRI for examinations in humans where motion is present. Feasibility was verified by examination in a first patient with glioblastoma.

Introduction

Chemical exchange-sensitive MRI after administration of glucose (glucoCEST/CESL) has been shown to provide insights into glucose perfusion/uptake, which is of particular interest for the identification of active tumor regions¹. Also its applicability for examinations in humans was already verified by two independent pilot studies^2,3. However, contrast maps are prone to motion-induced artifacts, as the final glucoCEST/CESL contrast relies on the difference of images which were acquired with a time gap of several tens of minutes. In this study, identification of different types of motion-induced artifacts (spatial displacement, spatially varying sensitivity of the coil, slab selection profile) led to the development of a 3D acquisition protocol for glucoCESL examinations in the human brain at 7T with improved robustness against motion.

Methods

To allow a clear identification of motion-induced artifacts one healthy volunteer was examined without administration of glucose. In addition, a first patient (glioblastoma, WHO grade IV) was examined with administration of D-glucose (2min,100ml,20%). The examinations were approved by the local ethics committee. In vivo 3D CESL MRI (1.7×1.7×3mm³,14 slices) was performed on a 7T-MR scanner (Siemens) using the snapshot-CEST⁴ approach. Image readout parameters were adapted from ref.4: 560 Hz/pix BW, Grappa 3, 6° FA. Pre-saturation was achieved by an adiabatically prepared spin-lock pulse that was optimized in previous studies^5,6(i.e. B$$$_1$$$=5µT, t$$$_{rec}$$$=5s). Images with and without R$$$_{1ρ}$$$ weighting (i.e. spin-lock time TSL=50 and 0.2ms) were acquired in an interleaved manner. Co-registration was performed separately for the two series of images with different TSL using a rigid registration algorithm in MITK⁷. The transformation information of the co-registration was used to quantify the presence of motion. B₁- and B₀-maps were calculated using the WASABI-method⁸. The dynamic glucose-enhanced (DGE$$$_\rho$$$)^3,6 contrast was calculated by$$$~DGE_\rho(t)=\frac{S_{50ms}(ref)\,-\,S_{50ms}(t)}{S_{50ms}(ref)\;\cdot\;50ms}~$$$[Eq.1], where S$$$_{50ms}$$$(t) are the acquired R$$$_{1ρ}$$$-weighted images, and S$$$_{50ms}$$$(ref) is the average of the first twenty S$$$_{50ms}$$$(t) images. The ΔR$$$_{1ρ}$$$ contrast based on quantitative R$$$_{1ρ}$$$ maps was calculated according to Jin et al.¹: $$$\Delta R_{1\rho}(t)=R_{1\rho}(t)-R_{1\rho}(ref)~$$$[Eq.2], with $$$R_{1\rho}(t)=\frac{\ln[S_{0.2ms}(t)/S_{50ms}(t)]}{50ms\;-\;0.2ms}~$$$[Eq.3], and R$$$_{1ρ}$$$(ref) being the average of the first ten R$$$_{1ρ}$$$(t) maps.

Results

The observed motion of the healthy volunteer during the approximately 40-minute-long examination was in the range of about ±0.5mm and ±1° in each spatial dimension (Fig.1e). Already, such subtle spatial displacements led to strong artifacts in the time course of the unregistered R$$$_{1ρ}$$$-weighted DGE$$$_\rho$$$ contrast (Fig.2a,b). A significant reduction of the motion-induced artifacts was observed after co-registration (Fig.2c,d). However, a global increase in the DGE$$$_{\rho}$$$ contrast was still present throughout almost the entire brain, originating mainly from the spatially varying sensitivity of the coil (Fig.1a). We found that an adequate suppression of this global artifact can be achieved by acquisition of a series of quantitative R$$$_{1ρ}$$$ maps (Fig.2e,f) instead of R$$$_{1ρ}$$$-weighted images. The variation of the resulting ΔR$$$_{1ρ}$$$ time course was in the approximate range of the noise level (i.e.±0.25Hz) and the center of the contrast histogram remained sufficiently constant over time (Fig.3). The final acquisition protocol, including co-registration and quantitative R$$$_{1ρ}$$$ mapping, was then utilized to investigate signal changes after administration of glucose in a patient with glioblastoma. A clear signal increase after glucose injection was observed on one side of the Gd-contrast ring-enhancement (Fig.4, white arrows), which did not correlate with observed B$$$_0$$$-inhomogeneities (Fig.4, black arrows).

Discussion

The advantage of quantitative ΔR$$$_{1ρ}$$$ imaging is due to the ratio of two consecutively acquired R$$$_{1ρ}$$$-weighted images (Eq.3). Assuming negligible motion during two consecutive images (Δt$$$~$$$≈$$$~$$$7s), this division normalizes the spatially varying sensitivity of the coil (compare Fig.1a,b) and slab selection profile, which are both dependent on the position of the head. In contrast, glucoCEST- and DGE$$$_\rho$$$-imaging is prone to changes in the spatially varying sensitivity of the coil or altered slab selection, as for normalization a reference image at the beginning of the examination is always used (Eq.1,$$$~$$$Δt$$$~$$$≈$$$~$$$up to several minutes). However, partial volume effects can still not be compensated completely by the quantitative ΔR$$$_{1ρ}$$$ approach which is why the interpretation of the ΔR$$$_{1ρ}$$$ contrast at sharp edges (e.g.tissue borders) has to be performed with caution. Nevertheless, in the examination of a tumor patient in this study, no negative signals were observed around the glucose-induced signal increase in the tumor region (Fig.4, white arrows), providing a preliminary verification of the applicability of the presented acquisition protocol to tumor imaging.

Conclusion

The presented 3D acquisition protocol allows glucoCESL examinations in the human brain with improved robustness against motion-induced artifacts. Removing artifacts for glucose-weighted MRI in humans is of high importance for future clinical trials in which an unequivocal assignment of contrast changes to an actual glucose perfusion/uptake is a prerequisite. A sufficient reduction of motion-induced artifacts may also allow transfer of glucose-weighted MRI to clinical magnetic field strengths (i.e.3T), where the contrast-to-noise ratio is even lower.

Acknowledgements

The financial support for MZ and KH of the Max Planck Society, German Research Foundation (DFG, grant ZA 814/2-1), and European Union’s Horizon 2020 research and innovation program (Grant Agreement No. 667510) is gratefully acknowledged.

References

1. Jin T, Iordanova B, Hitchens TK, et al. Chemical exchange–sensitive spin-lock (CESL) MRI of glucose and analogs in brain tumors. Magn Reson Med 2018; 80:488-495.

2. Xu X, Yadav NN, Knutsson L, et al. Dynamic Glucose-Enhanced (DGE) MRI: Translation to Human Scanning and First Results in Glioma Patients. Tomography 2015; 1(2):105-114.

3. Paech D, Schuenke P, Koehler C, et al. T1ρ-weighted Dynamic Glucose-enhanced MR Imaging in the Human Brain. Radiology 2017; 285(3):914-922.

4. Zaiss M, Ehses P, and Scheffler K. Snapshot-CEST: Optimizing spiral-centric-reordered gradient echo acquisition for fast and robust 3D CEST MRI at 9.4 T. NMR Biomed 2018; 31:e3879.

5. Schuenke P, Koehler C, Korzowski A, et al. Adiabatically Prepared Spin-Lock Approach for T1ρ-Based Dynamic Glucose Enhanced MRI at Ultrahigh Fields. Magn Reson Med 2017; 78:215-225.

6. Schuenke P, Paech D, Koehler C, et al. Fast and Quantitative T1ρ-weighted Dynamic Glucose Enhanced MRI. Scientific Reports 2017; 7: 2045-2322.

7. Nolden M, Zelzer S, Seitel A, et al. The Medical Imaging Interaction Toolkit: challenges and advances. Int J CARS 2013; 8(4):607-620.

8. Schuenke P, Windschuh J, Roeloffs V, et al. Simultaneous mapping of water shift and B1(WASABI)-Application to field‐Inhomogeneity correction of CEST MRI data. Magn Reson Med 2017, 77: 571-580.

Figures

Fig. 1: (a) R_1ρ-weighted image of a healthy volunteer (slice 2/14, TSL=50ms) showing strong signal variations dependent on the spatially varying sensitivity of the coil. The specified regions of interest (ROI) were used for calculation of the illustrated time-courses in Fig.2. (b) Quantitative R_1ρ-map calculated from two consecutive images (Eq.3) showing homogeneous contrast independent of field-inhomogeneities (c,d), verifying the functionality of the adiabatic spin-lock preparation. (e) Time-course of motion during the approximately 40-minute-long examination (please note the three distinct parts in which the volunteer was asked not to move, to relax, and to slightly move the head in one direction).

Fig. 2: Time courses of different glucoCESL contrasts based on (a-d) R_1ρ-weighted images, and (e, f) quantitative R_1ρ maps (not averaged). The signal values of the different contrasts are directly comparable due to calculation of DGE_ρ in the unit Hz. The ROIs (b, d, f) for calculation of the illustrated time courses are specified in Fig. 1. Notice the motion-induced artifacts at later time points in (a-d) merely arising from sub-mm movement. (e, f) Quantitative ΔR_1ρ mapping in combination with a 3D co-registration enabled glucoCESL MRI in the human brain with an adequate suppression of motion-induced artifacts.

Fig. 3: Contrast histograms of the co-registered (a) DGE_ρ, and (b) ΔR_1ρ maps illustrated in Fig. 2c, e. The histograms depict the three distinct parts in which the volunteer was asked not to move (blue), to relax (green), and to slightly move the head in one direction (orange). The centers of the co-registered DGE_ρ histograms shift due to a false, motion-induced signal increase originating mainly from the spatially varying sensitivity of the coil (Fig. 1a). In contrast, the centers of the co-registered ΔR_1ρ histograms remain approximately constant, thus illustrating the robustness of the presented acquisition protocol against motion-induced artifacts.

Fig. 4: (a) Quantitative ΔR_1ρ mapping of a patient with glioblastoma (slice 8-10/14, average of eleven quantitative ΔR_1ρ maps) showing a signal increase on one side of the Gd-contrast ring-enhancement after administration of glucose (white arrows). No negative signals in neighboring voxels occur in any of the three dimensions, thus verifying a glucose-induced signal increase independent of motion-induced artifacts. In addition, no correlation of the signal time course (b) with the patient motion (c) is present. Regions of strong B₀-inhomogeneities (black arrows) cannot be evaluated reliably. Please note that the evaluated blood vessel ROI is outside of the illustrated slices.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)

5009