3111

Initial experience of imaging stroke patients with a 5-minute whole head multi-echo GRE acquisition
Junmin Liu1, Luciano Sposato2, Spencer D Christiansen1,3, and Maria Drangova1,3

1Imaging Research Laboratories, Robarts Research Institute, Schulich School of Medicine & Dentistry, University of Western Ontario, London, Ontario, Canada, London, ON, Canada, 2Department of Clinical Neurological Sciences, Schulich School of Medicine & Dentistry, University of Western Ontario, London, Ontario, Canada, London, ON, Canada, 3Department of Medical Physics, Schulich School of Medicine & Dentistry, University of Western Ontario, London, Ontario, Canada, London, ON, Canada

Synopsis

We report on our initial experience of imaging stroke patients using a technique that achieves quantification of fat fraction (FF), QSM, and R2* simultaneously from a single multi-echo GRE (mGRE) acquisition. We collected 3T data from three stroke patients with a ~ 30-minute whole head multi-sequence protocol. By performing the joint analysis of TOF MRA, DWI, FLAIR and mGRE, we evaluated the capability of using the mGRE-based maps and images to characterize thrombus, differentiate intracranial calcifications from hemorrhages and detect white-matter lesions. Our initial results have shown the feasibility of using the mGRE technique to image stroke patients.

INTRODUCTION

T2*-weighted imaging techniques have demonstrated comparable accuracy with CT for detecting parenchymal hemorrhage.1 Recently these susceptibility-sensitive techniques (including SWI 2 ) have also shown potential for qualitative characterization of thrombus, i.e., identifying stroke subtype by detecting “susceptibility vessel sign (SVS)” 3. Conventionally, these techniques collect data from a single-echo GRE acquisition with long echo times. However, quantitative characterization (e.g., R2* map) of hemorrhage and thrombus is impossible from a single-echo acquisition, necessitating the use of a multi-echo GRE (mGRE) approach. Furthermore, since most thrombi of interest are located in areas of relatively severe B0 inhomogeneities and co-existence of fat and water (brain near the base of the skull and neck) we expect that accurate proton density fat-fraction (FF) and B0 mapping will benefit the quantitative mapping and visualization.

We have previously described a technique 4, which is capable of performing Dixon MRI (i.e., fat-water separation), multi-contrast imaging and multi-parametric mapping from a single bipolar dual echo-train mGRE acquisition. In this work, we report our initial experience of imaging stroke patients and demonstrate the utility of imaging stroke patients using the described technique.


METHODS

Protocol and data acquisition: Data from 3 stroke patients were acquired on a 3T scanner (Siemens Prisma) with a ~30-minute whole head multi-sequence stroke protocol, including TOF MRA, DWI, FLAIR, with an additional mGRE 4. Specifically, a bipolar dual echo-train mGRE acquisition was prescribed to cover the whole head in approximately 5 minutes: the first echo train (five echoes) was optimized for fat/water separation (TE = 3.3, 4.7, 6.2, 7.7 and 9.5 ms), while the second echo train was designed with TEs optimized for susceptibility mapping with fat and water in-phase (16.8, 23.9, 31.1, 38.2, and 45.4 ms). The remaining scan parameters were: FA 15°; TR 51 ms; bandwidth 1015 Hz/pixel; spatial resolution 1.0×1.0×2.0 mm3; GRAPPA = 2.

mGRE Processing: The detailed description of the processing algorithm was presented in a previous publication 4 : First, we corrected the phase errors associated with the bipolar acquisition; Second, we used the unwrapping-based B0 mapping technique to map FF and B0 5 ; Third, we generated a total field map using a complex fitting method and then used the predetermined B0 map for background removal, followed by calculating the QSM map using the morphology enabled dipole inversion algorithm 6 ; Finally, R2* maps were generated from all ten echoes using exponential fitting. We generated anatomical navigation images by averaging the magnitude images over the first echo train. Note that only theseven in-phase echoes were used in QSM quantification and the reported QSM values were calculated relative to the mean susceptibility of the central CSF region in the lateral ventricles.

Image analysis: Patient etiology was determined from prior CT and MR scans. We aimed to evaluate the capability of using the mGRE-based maps and images to characterize thepatient’s thrombus etiology(hyperintense on R2* and hypointense on TOF) 7 , differentiate intracranial calcifications (hypointense on QSM and hyperintense on R2*) from hemorrhages (hyperintense on both R2* and QSM) 8 , and detect white matter lesions (hypointense on R2*, and hyperintense on the DWI and FLAIR). For quantitative characterization, thrombi were manually segmented on the R2* map and the mean R2* and QSM were calculated.

RESULTS and DISCUSSION

In Fig. 1, we show the anatomical navigation images, FF, R2* and QSM of patient #1 with an M2/M3 occlusion (arrows). This figure demonstrates the ability to detect and quantify the parameters in smaller vessels, despite the relatively large voxel sizes (1x1x2 mm3). We also show the corresponding TOF (Fig. 1e), DWI (Fig. 1f) and FLAIR (Fig. 1g) images, which are representative of the current standard of imaging. In Fig. 2 we show the results from patient #2 with a thrombus in the left vertebral artery – an extremely challenging area to image with suscetibility-based MRI due to its location near the base of the skull and top of neck, where the field inhomogeneities are large. In Fig. 3, we show results from patient # 3, who was imaged because of a disection of the right carotid; the figure shows an incidental findinding of a calcified choroid plexus, which demosntrates the capability of the whole-head QSM to identify intracranial calcifications. Of additional interest is the ability to detect white matter lesions on the R2* maps (demonstrated in Fig. 4), which coincide with those seen on FLAIR and DWI.

CONCLUSION

Our initial results show the feasibility of using dual-echo train mGRE technique to image stroke patients and suggest the potential of providing quantitative measures of thrombus and infarction, which needs to be demonstrated with larger patient cohorts.

Acknowledgements

Partial funding was provided by a grant from the Natural Sciences and Engineering Research Council of Canada and a grant from Canadian Institutes of Health Research (PJT 153411).

References

1. Baird AE, Warach S. Magnetic resonance imaging of acute stroke. Journal of cerebral blood flow and metabolism : official journal of the International Society of Cerebral Blood Flow and Metabolism 1998;18(6):583-609.

2. Haacke EM, Xu Y, Cheng YC, Reichenbach JR. Susceptibility weightedimaging (SWI). Magn Reson Med 2004;52(3):612-618..

3. Kang DW, Jeong HG, Kim DY, Yang W, Lee SH. Prediction of Stroke Subtype and Recanalization Using Susceptibility Vessel Sign on Susceptibility-Weighted Magnetic Resonance Imaging. Stroke 2017;48(6):1554-1559.

4. Liu J, Christiansen SD, Drangova M. Single multi-echo GRE acquisition with short and long echo spacing for simultaneous quantitative mapping of fat fraction, B0 inhomogeneity, and susceptibility. Neuroimage 2018;172:703-717.

5. Liu J, Drangova M. Method for B0 off-resonance mapping by non-iterative correction of phase-errors (B0-NICE). Magn Reson Med 2015;74(4):1177-1188.

6. Liu T, Liu J, de Rochefort L, et al. Morphology enabled dipole inversion (MEDI) from a single-angle acquisition: comparison with COSMOS in human brain imaging. Magn Reson Med 2011;66(3):777-783.

7. Bourcier R, Brecheteau N, Costalat V, et al. MRI quantitative T2* mapping on thrombus to predict recanalization after endovascular treatment for acute anterior ischemic stroke. Journal of neuroradiology Journal de neuroradiologie 2017;44(4):241-246.

8. Chen W, Zhu W, Kovanlikaya I, et al. Intracranial calcifications and hemorrhages: characterization with quantitative susceptibility mapping. Radiology 2014;270(2):496-505.

Figures

Figure 1. Results from patient #1 with an M2/M3 occlusion (arrows), thrombus appearance shows: SVS on the navigation images (a); hyperintensity on the FF (b), R2* (c) and QSM maps (d); and hypointensity on TOF (e). The corresponding infarcts appear as hyperintensity on DWI (f) and FLAIR (g). For the identified thrombus, R2* = 157 ±70 s-1, susceptibility = 0.30 ±0.32 ppm. Noticeably, the size of the area with hyperintensity on FF was smaller than that on R2* and QSM; this observation could correspond to the existence of lipid core or artifact introduced by rapid signal decay.

Figure 2. Patient #2 with a thrombus (orange arrows) in the left vertebral artery (V3/V4) and with posterior fossa infarct (purple arrows). After supressing fat signal on the R2* map (multiplying (c) by (1-FF)), we observed improved visualization of thrombus and infarct. The hypointensity on R2* corresponds well to the hyperintensity on FLAIR. However, the green arrows identify an area with hyperintensity on R2* (e) that is seen as normal tissue on FLAIR (f), suggesting that R2* could provide important additional information about the infarct. For the R2* maps, the display range is from 0 to 80 s-1.

Figure 3. Images from patient #3 identify a choroid plexus calcification. The arrows point to: hyperintensity on the R2* map (a); hypointensity on QSM(b); hyperintensity on the FLAIR (c); and hypointensity on the DWI images (d). Among these image contrasts, the QSM image distinctly shows a diamagnetic lesion, which corresponds to choroid plexus calcification. While this finding is not related to stroke, it clearly demonstrates the capacity of the QSM/ R2* combination to identify calcified lesions.

Figure 4. Examples of white matter lesions (arrows) identified in: (a) patient #1; (b) patient #2; (c) patient #3. Overall, the total number of R2* lesions is lower than the that seen in the FLAIR or DWI images and the average size of R2* lesion is larger that that of the corresponding FLAIR lesion. For the R2* maps, the display range is from 0 to 80 s-1.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)
3111