1172

Geometric Distortions and Signal-To-Noise Ratio of Conventional and Inner Fields-of-View for T2*-Weighted Echo-Planar Imaging of the Spinal Cord
Ying Chu1 and Jürgen Finsterbusch1

1Department of Systems Neuroscience, University Medical Center Hamburg-Eppendorf, Hamburg, Germany

Synopsis

Geometric distortions and signal-to-noise ratios (SNR) of T2*-weighted echo-planar imaging (EPI) of the spinal cord are compared for conventional and inner-FOV acquisitions based on 2D-selective RF (2DRF) excitations. For conventional acquisitions, the required FOV increases with the in-plane object size yielding more pronounced distortions, prolonged echo times (TEs), and reduced SNR. For inner-FOV acquisitions, the FOV is small and independent of the object size yielding only minor distortions. The 2DRF pulse duration must be adapted for larger object sizes resulting in slightly prolonged TEs but overall TEs remain shorter and SNR values are larger than for conventional acquisitions.

Introduction

Echo-planar imaging (EPI)1 of inner-fields-of-view (FOV) based on spatially Two-dimensional selective RF (2DRF) excitations2,3 allows to focus the FOV to an inner target region without aliasing4 which can reduce geometric distortions and the echo time (TE) and, thus, increase the image quality and signal-to-noise ratio (SNR). It has been applied successfully for diffusion-weighted imaging of the spinal cord,5,6 and T2*-weighted imaging of small brain regions,7 however, its potential for blood-oxygenation-level-dependent (BOLD) functional neuroimaging of the spinal cord8 is still under research. In this study, conventional and 2DRF-based inner-FOV T2*-weighted EPI acquisitions are compared and evaluated concerning geometric distortions and SNR for different in-plane object sizes.

Methods

The basic EPI pulse sequences used for conventional slice-selective and inner-FOV acquisitions based on 2DRF excitations are shown in Fig. 1.

The geometric setups for acquisitions of inner targets like the spinal cord are sketched in Fig. 2. For conventional acquisitions, the required phase-encoding FOV depends on the in-plane body or object size in the phase encoding direction. Some aliasing may be tolerable as long as parallel acquisition techniques (PAT) are not used and the desired target region is not affected (Fig. 2a). For inner-FOV acquisitions, the FOV only has to cover the target region, i.e. is independent of the object size, but the field-of-excitation (FOE) of the 2DRF pulse must be large enough to position the side excitations outside of the object to avoid unwanted signal contributions. (Fig. 2b) This means that the FOE and, thus, the 2DRF pulse length depends on the in-plane object size.

Measurements were performed on a 3T whole-body MR-system (PrismaFit, Siemens Healthineers) using a 20-channel head coil for the phantom and the neck coil elements of a 64-channel head-neck coil for in vivo acquisitions. Healthy volunteers were investigated after their informed consent was obtained. All acquisitions were performed using an in-plane resolution of 1.0×1.0mm2 and a slice thickness of 5mm with a fixed repetition time (TR) of 4030ms (in vivo) and 3000ms (phantom). Protocols for a target size (inner-FOV) of 32mm and in-plane object sizes between 160mm (cervical spinal cord) and 416mm (thoracic/lumbar spinal cord in obese patients) were considered yielding FOVs between 96mm and 224mm for conventional acquisitions and FOEs between 110mm and 250mm for inner-FOV acquisitions (2DRF pulse durations between 11.5ms and 26.0ms). Conventional acquisitions were performed without and with PAT because the receive coils, which are available for the spinal cord may not be feasible for PAT in lower cord sections. Thus, conventional EPI was performed with TEs between 23ms and 46ms (PAT) and, 41ms and 86ms (no PAT). Inner-FOV EPI was performed with 40% phase-encoding oversampling to consider imperfections of the 2DRF excitation profile (resolution 2.5×10.0mm2) yielding TEs between 26ms and 33ms.

All in vivo tests (12 slices) were performed in the cervical spinal cord for a direct comparison of distortions and SNR for different object sizes.

SNR values in the spinal cord were determined using a self-written IDL procedure by taking the voxel-wise ratio of the mean and standard deviation of 20 measurements and averaging this ratio over the spinal cord cross-section in each slice.

Result and Discussion

The images in Fig. 3 demonstrate that the geometric distortions are more pronounced in conventional than in inner-FOV acquisitions and increase with the object size due to the larger FOV required. In vivo images of a healthy volunteer are shown in Fig. 4. As in the phantom acquisitions, increased distortions are visible for conventional acquisitions together with a significant SNR reduction for the large object size due to the much longer TE required.

Figure 5 presents the SNR values in a single slice and averaged over 10 slices as a function of the object size. Inner-FOV acquisitions show a good SNR with only a slight decrease with the object size due to an increased FOE and a correspondingly prolonged 2DRF pulse and TE. Conventional acquisitions without PAT have a lower SNR and show a stronger decrease with the object size due to the increased FOV and TE which are longer than inner FOV acquisitions. With PAT, conventional acquisitions show an improved SNR for large object sizes due to the shorter echo train and TE but still have a lower SNR compared to inner-FOV acquisitions. For smaller object sizes, PAT acquisitions show a bad performance because aliasing artifacts interfere with PAT.

Conclusion

In conclusion, inner-FOV acquisitions provide reduced geometric distortions and an improved SNR for T2*-weighted imaging of the spinal cord, in particular for lower cord sections and obese patients, and could help to enhance BOLD-based neuroimaging.

Acknowledgements

No acknowledgement found.

References

1. Mansfield P. Multi-planar image formation using NMR spin echos. J Phys C. 1977; 10: 55-58

2. Bottomley PA, Hardy CJ. Two-dimensional spatially selective spin inversion and spin-echo refocusing with a single nuclear magnetic resonance pulse. J Appl Phys. 1987; 62: 4284-4290.

3. Pauly J, Nishimura D, Macovski A. A k-space analysis of small-tip-angle excitations. J Magn Reson. 1989; 81: 43-56.

4. Rieseberg S, Frahm J, Finsterbusch J, Two-dimensional spatially-selective RF excitation Pulses in echo-planar imaging. Magn Reson Med. 2002; 47: 1186-1193

5. Saritas EU, Cunningham CH, Lee JH, Han ET, Nishimura DG. DWI of the spinal cord with reduced FOV single-shot EPI. Magn Reson Med. 2008; 60: 468-473.

6. Finsterbusch J. High-resolution diffusion tensor imaging with inner field-of-view EPI. J Magn Reson Imaging. 2009; 29: 987-993

7. Finsterbusch J. Functional neuroimaging of inner fields-of-view with 2D-selective RF excitations. Magn Reson Imaging. 2013; 31: 1228–1235

8. Eippert F, Finsterbusch J, Bingel U, Büchel C. Direct Evidence for Spinal Cord Involvement in Placebo Analgesia. Science. 2009; 326: 404

9. Oelhafen M, Pruessmann KP, Kozerke S, Boesiger P. Calibration of Echo-Planar 2D-selective RF excitation pulses. Magn Reson Med. 2004; 52: 1136-1145

Figures

Fig. 1: Basic pulse sequences used in the current study for (a) conventional and (b) inner-FOV EPI acquisitions. For inner-FOV EPI, the initial slice-selective RF excitation of the conventional EPI sequence is replaced by a 2DRF excitation based on a fly-back blipped-planar trajectory9 with the line and blip direction coinciding with the imaging slice and phase direction, respectively. For simplicity, only a few lines of the echo train and of the 2DRF trajectory are sketched.

Fig. 2: Geometric setups for slices at different body cross-sections for (a) conventional and (b) inner-FOV acquisitions with the filled boxes representing the target region. For conventional acquisitions, the smallest FOV without aliasing into the target region (dashed boxes) increases with the in-plane object size. For inner-FOV acquisitions, a fixed FOV can be used for all object sizes but the field-of-excitation (FOE) increases with the object size to position the side excitations (empty boxes) outside of the body to avoid unwanted signal contributions. The FOE was rounded to odd multiples of the resolution of the 2DRF trajectory (10mm).

Fig. 3: Measurements of a kohlrabi with a plastic cylindrical wall in the middle for different object sizes demonstrating the geometric distortions in the presence of field inhomogeneities as induced by a slight misadjustment of the shim setting in the phase-encoding direction. (a) fast low-angle shot (FLASH) acquisition without geometric distortions showing a cylinder diameter of 30mm, (b) conventional acquisition without PAT (cylinder diameter for small/large (top/bottom) object size 27mm/24mm), (c) conventional acquisition with PAT (28mm/27mm), and (d) inner-FOV acquisition (29mm/29mm). The distortions are more pronounced for conventional acquisitions and, in contrast to inner-FOV acquisitions, increase with the object size.

Fig. 4: One slice of an in vivo acquisition for different in-plane object sizes. For conventional acquisitions, the FOV must be increased for larger object sizes which amplify geometric distortions, prolongs the echo time (without/with PAT from 52/29ms to 86/46ms), and reduces the signal-to-noise ratio accordingly. For inner-FOV acquisitions, a fixed FOV can be used for all object sizes, only the field-of-excitation must be adapted which, however, does not affect the geometric distortions and only yields a minor prolongation of TE (from 28ms to 33ms) and SNR reduction.

Fig. 5: Plots of SNR values in the spinal cord for (a) a single slice, which is partly shown in Fig. 4 (b) The average of all 10 slices analyzed for object sizes between 160mm and 416mm. Two slices had to be discarded because the geometric distortions in some of the conventional acquisitions without PAT were too severe to obtain reliable values.There is one missed data point of inner FOV because the measurements are missing. However, the SNRs of inner-FOV near the data points are similar, so the missing of a data point is not significant for the results.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)
1172