1902

Evaluation of a Novel 8-Channel Receive Coil for Speech Production MRI at 0.55 Tesla.

Felix Munoz¹, Yongwan Lim², Sophia X. Cui³, Helmut Stark⁴, and Krishna Shrinivas Nayak^1,2
¹Department of Biomedical Engineering, University of Southern California, Los Angeles, CA, United States, ²Ming Hsieh Department of Electrical and Computer Engineering, University of Southern California, Los Angeles, CA, United States, ³Siemens Medical Solutions, Los Angeles, CA, United States, ⁴Stark Contrast, Erlangen, Germany

Synopsis

MRI has enabled significant advances in our understanding of human speech production. This application benefits from a lower magnetic field due to reduced off-resonance effects at air-tissue interfaces. Here, we evaluate the SNR and parallel imaging performance of a dedicated speech coil at 0.55 Tesla, and compare it to that of a head-neck coil. Over the upper airway regions of interest, the dedicated coil shows approximately 1.3- to 4.6-fold improvement in SNR efficiency compared to the head-neck coil.

Introduction

MRI is a powerful tool to visualize the complex coordination of articulators, enabling advances in speech science and clinical applications¹. The linear scaling of susceptibility variation with field strength makes low field beneficial, especially because the features of interest are at the air-tissue interfaces². Multi-channel local coils also provide increased SNR, and the ability to perform parallel imaging, increasing scan efficiency³.

In this work, we evaluate a custom upper airway (UA) coil for 1H imaging at 0.55 Tesla (23.6 MHz) and evaluate its performance in comparison with a head/neck (HN) coil over nine upper airway volumes-of-interest (VOIs). We also evaluate the parallel imaging performance for the dedicated UA coil using Cartesian SENSE with uniform and variable density undersampling. Similar studies at 1.5 and 3 Tesla showed significant SNR improvement with localized speech coils (up to 5-fold) ^4-5.

Methods

Experimental Methods: Experiments were performed using a whole body 0.55T system (prototype MAGNETOM Aera, Siemens Healthineers, Erlangen, Germany) equipped with high-performance shielded gradients (45 mT/m amplitude, 200 T/m/s slew rate). The dedicated 8 channel UA coil (Stark Contrast, Germany) and the standard HN coil with 16 channels are shown in Figure 1. 3D volumes of the upper airway were acquired using 3D gradient echo imaging. Imaging parameters were: flip angle = 10^o, TE = 5 ms, TR = 10 ms, FOV = 32x32x16 cm³, resolution = 2.5 x 2.5 x 5 mm³, receiver bandwidth = 150 Hz/Px, and k_y and k_z phase encoding along A-P and R-L directions, respectively. Four subjects were scanned in a supine position under a protocol approved by our Institutional Review Board, after providing written informed consent.
Data Analysis: Pre-scan noise was acquired with 256*128 samples per channel and used for pre-whitening. SNR maps were computed with the following equation⁶:
$$ SNR =\frac{ \sqrt{2} \left| \bf{b}^{H}\bf{p} \right|} {\sqrt{\left( \bf{b}^{H} \bf{b} \right) } } $$
where p is the vector of complex image values for each coil, b is the vector of complex coil sensitivities. Image and coil sensitivity data were pre-whitened prior to SNR evaluation. To compare the SNR efficiency of both coils, each coil was first compared to the integrated body coil by performing the imaging sequence without patient repositioning. This allowed calculation of SNR gains for each of UV and HN coils relative to the body coil. Subsequently, the ratio of SNR gains for each coil was obtained.

Figure 2 illustrates the 9 VOIs used for SNR evaluation. Each VOI was edited using the 3D segmentation tools of ITK-SNAP⁷. For parallel imaging evaluation, g-factors were calculated for conjugate gradient (CG) SENSE with uniform density undersampling (UD), and for variable density (VD) Cartesian trajectory using 32 central lines. We tested acceleration factors of 2, 3, and 4. For CG SENSE, the g-factor at pixel ρ is defined as⁸:
$$ g_{\rho} = \sqrt{(E^{H}E)^{-1}_{\rho,\rho} \: (E^{H}E)_{\rho,\rho}} $$
where E is the encoding matrix as defined in [9].

Results

Figure 3 compares relative SNR for the UA and HN coil in all four subjects. The UA coil provides higher SNR gain in the upper airway regions whereas the HN coil provides higher SNR gain in the brain regions, as expected. This is supported by the quantitative measure of SNR gain as shown in Figure 4. The largest gains were in the upper and lower lip areas. The HN coil outperforms the UA coil in regions closer to the brain, such as the frontal lobe.

Figure 5 contains g-factor maps, along with reconstructed images. For R = 2 and 3, both UD and VD trajectories have g-factor below 1.1 in the upper airway region. For R = 4, the g-factor are higher for both sampling schemes.

Discussion

The placement of coil elements allows for a high SNR gain near the vocal tract articulators, while the gain is negligible near the pharyngeal wall. The high SNR observed near the articulators is beneficial to upper airway imaging, as it mitigates g-factor noise amplification allowing for better image quality¹⁰.

Increasing SNR by minimizing coil size is limited by coil loses. In the present coil element size is optimized for deeper structures. The quality factor Q₀ is optimized and the highest loading is achieved by anatomical shaping to keep distance to the skin small. For even faster imaging, a design with more elements can help acceleration (if g-factors allow). This could give higher SNR near the surface. With a tradeoff for loading factor at this low frequency, coil losses contributions will increase significantly and SNR at depth will decrease. Inductive coupling between elements at low frequency will also increase noise: this effect can be worse for more elements.

Conclusion

The novel UA coil showed a 1.3-4.6-fold improvement in SNR over the HN coil in VOIs relevant to speech production. The g-values for the UA coil in the upper airway regions are below 1.1 for R = 2 and 3, for uniform and variable density CG SENSE. This is favorable for high-performance speech production dynamic MRI at 0.55 Tesla.

Acknowledgements

We acknowledge grant support from the National Science Foundation (#1828736) and research support from Siemens Healthineers

References

1. Lingala SG, Sutton BP, Miquel ME, Nayak KS. Recommendations for real-time speech MRI. Journal of Magnetic Resonance Imaging. 2016;43(1):28–44. doi:10.1002/jmri.24997

2. Bernstein MA, Huston J, Ward HA. Imaging artifacts at 3.0T. Journal of Magnetic Resonance Imaging. 2006;24(4):735–746. doi:10.1002/jmri.20698

3. Roemer PB, Edelstein WA, Hayes CE, Souza SP, Mueller OM. The NMR phased array. Magnetic Resonance in Medicine. 1990;16(2):192–225. doi:https://doi.org/10.1002/mrm.1910160203

4. Kim Y-C, Hayes CE, Narayanan SS, Nayak KS. Novel 16-channel receive coil array for accelerated upper airway MRI at 3 Tesla. Magnetic resonance in medicine. 2011;65(6). doi:10.1002/mrm.22742

5. Lingala SG, Zhu Y, Kim Y, Toutios A, Narayanan S, Nayak KS. A fast and flexible MRI system for the study of dynamic vocal tract shaping. Magnetic Resonance in Medicine. 2017;77(1). doi:10.1002/mrm.26090

6. Kellman P, McVeigh ER. Image reconstruction in SNR units: A general method for SNR measurement. Magnetic Resonance in Medicine. 2005;54(6):1439–1447. doi:10.1002/mrm.20713

7. Yushkevich PA, Piven J, Hazlett HC, Smith RG, Ho S, Gee JC, Gerig G. User-guided 3D active contour segmentation of anatomical structures: Significantly improved efficiency and reliability. NeuroImage. 2006;31(3):1116–1128. doi:10.1016/j.neuroimage.2006.01.015

8. Sheng J, Ying L, Liu B, Abdelsalam E, Sheng J, Ying L. G-factor Maps of Conjugate Gradient SENSE Reconstruction. 2008.

9. Pruessmann KP, Weiger M, Börnert P, Boesiger P. Advances in sensitivity encoding with arbitrary k-space trajectories. Magnetic Resonance in Medicine. 2001;46(4):638-651. doi:10.1002/mrm.1241

10. De Zwart JA, Ledden PJ, Van Gelderen P, Bodurka J, Chu R, Duyn JH. Signal-to-Noise Ratio and Parallel Imaging Performance of a 16-Channel Receive-only Brain Coil Array at 3.0 Tesla. Magnetic Resonance in Medicine. 2004;51(1):22–26. doi:10.1002/mrm.10678

Figures

Figure 1: Placement of UA and HN coil on an adult human subject. (left) 8-Channel UA coil. Note the proximity of the coil elements to the upper airway region. A 9th element can be placed in between the elements for increased SNR, if the subject’s head size allows. (right) 16-Channel HN coil. Notice that the UA coil is closer to the vocal tract articulators of interest and excludes signal and noise from the brain.

Figure 2: Upper airway volumes of interest (VOIs) used for evaluation. Top row: Intersection of the VOIs with a mid-sagittal slice acquired using the (left) UA coil and (right) HN coil. Bottom row: Volume rendering of 3D VOIs. Corresponding VOIs have similar volumes.

Figure 3: SNR relative to the body coil $$$\frac{SNR_{coil}}{SNR_{body}}$$$ for 4 subjects. In all subjects, the UA coil provided a SNR gain in the upper airway VOIs (white arrow). The HN coil provided higher SNR in the brain (red arrow), as expected.

Figure 4: Relative SNR of the UA coil compared to the HN coil. The height and color of the bars represent relative SNR. The UA coil outperforms the HN coil in vocal tract VOIs. The biggest improvement is in the upper and lower lip with 4.3-fold relative SNR.

Figure 5: Parallel imaging evaluation in the mid-sagittal slice with anterior-posterior undersampling direction. First and third row: UD CG and VD CG reconstructions, respectively. Second and fourth row: UD CG and VD CG gmaps, respectively. For R=2 and 3, g-values average about 1.1 in the upper airway region for both trajectories, allowing for good image quality for speech imaging. At R=4, g-values are higher (about 1.6), resulting in a more noisy reconstruction.

Proc. Intl. Soc. Mag. Reson. Med. 30 (2022)

1902

DOI: https://doi.org/10.58530/2022/1902