Accelerated Multi Echo based Correlated spectroscopic imaging of calf muscle in three spatial dimensions
Manoj Kumar Sarma1, Zohaib Iqbal1, Rajakumar Nagarajan1, and M. Albert Thomas1

1Radiological Sciences, UCLA School of Medicine, Los angeles, Los Angeles, CA, United States

Synopsis

Multi-echo based echo-planar correlated spectroscopic imaging (ME-EP-COSI) has been an innovative method to study muscle lipid content in T2D and a variety of other metabolic conditions. In this study we implemented accelerated ME-EP-COSI and validated in a corn oil phantom and in healthy human calf muscle. Both phantom and human calf muscle results show that 5D ME-EP-COSI has the potential to be a powerful tool for human calf muscle examination. Further studies will investigate various pathologies, including obesity and type 2 diabetes, using the 5D ME-EP-COSI method.

Purpose/Introduction:

One dimensional magnetic resonance spectroscopy (1D MRS) suffers from severe spectral overlap, making it difficult to quantify overlapping resonances. In order to overcome this disadvantage, Localized correlated spectroscopy (L-COSY)1 spreads signal over a second spectral dimension. Furthermore, combining L-COSY with spectroscopic imaging and utilizing an echo-planar readout leads to the echo-planar correlated spectroscopic imaging (EP-COSI)2 sequence, which acquires two dimensional spectra from multiple spatial regions. This type of acquisition takes an enormous amount of time, even with the use of Multi-echo (ME) based methods3. Recently, non-uniform sampling (NUS) and compressed sensing (CS)4 reconstruction have been applied to accelerate the EP-COSI method to acquire five dimensional (3 spatial and 2 spectral dimensions) EP-COSI data5 in human calf muscle, where intramyocellular (IMCL) and extramyocellular (EMCL) lipids are well documented. In this study, a novel approach incorporating ME acquisition, NUS and CS reconstruction, called 5D ME-EP-COSI, was validated in a corn oil phantom and in healthy human calf muscle.

Materials and Methods:

The standard 5D ME-EP-COSI sequence which uses a 90°–180°-Δt1-180° scheme for localization was modified by imposing non-uniform under-sampling (4X) along the t1 and z-direction (Figure 1). The quality of the CS reconstructed NUS 5D ME-EP-COSI data were compared using fully-encoded and prospectively undersampled phantom scans. A corn oil phantom was used for acquiring 10 in vitro measurements. The sequence was tested in the calf muscle of five healthy volunteers (age of 23-58 years). All data were collected on a 3T Prisma MRI scanner using a 15 channel knee ‘receive’ coil. The following parameters were used for both fully sampled and NUS- based ME-EP-COSI phantom data: TR/TE = 2s/30 ms, voxel resolution=3.37cm3, 64 Δt1 increments, 256 bipolar echo pair, FOV= 24x24x12 cm2, F1 and F2 bandwidths of 1250 Hz and 1190 Hz respectively. A non-water-suppressed ME-EP-COSI data with t1=1 were also recorded for eddy current correction and coil combination. For in-vivo NUS data TR was 1.2s and voxel resolution=1.5cm3 with scan time ~21min. Other scan parameters were the same as phantom. A skewed squared sine-bell sampling density scheme was used for both the phantom and in-vivo studies. The data were reconstructed6 using $$\min_{u} TV(u) \quad \text{s.t. } \|R\mathcal{F}u - f\|_2^2 < \sigma^2$$ where u is the reconstructed 5D data, TV is total variation, R is the sampling mask, $$$\mathcal{F}$$$ is the Fourier transformation along the non-uniformly sampled dimensions, f is the sampled data, and $$$\sigma$$$ is an estimate of the noise variance. Acquired 5D data were post-processed with a custom MATLAB-based program first to sort out the two EPSI read-out trains and then reconstructed using CS along the z and t1 dimensions.

Results and Discussion:

Figure 1 shows the 5D ME-EP-COSI pulse sequence diagram. The NUS phase-encoding dimensions (kz,t1) were sampled according to the mask shown in Figure 2. Sampled points are shown in red, whereas points that are not sampled are shown in black. A comparison between the fully sampled (A), NUS (B), and reconstructed (C) corn oil phantom is shown in Figure 3. As seen from the figure, ridging and other artifacts are removed post reconstruction. In vivo human calf muscle spectra can be seen in Figure 4 from both the soleus muscle (B) and the marrow (C). The absence of the Cr3.9 peak in the marrow was expected and shows that the spatial information is preserved using the accelerated ME-EP-COSI method. The methyl fat spatial profile for three slices is shown in Figure 4A.

Conclusion:

We presented here a 5D multi-echo-based correlated spectroscopic imaging sequence with echo planar readout which is only feasible by applying NUS along one phase-encoding and one indirect spectral dimension. ME-EP-COSI have been an innovative method to study muscle lipid content in T2D and a variety of other metabolic conditions3. Both phantom and human calf muscle results show that 5D ME-EP-COSI has the potential to be a powerful tool for human calf muscle examination. Further studies will investigate various pathologies, including obesity and type 2 diabetes, using the 5D ME-EP-COSI method.

Acknowledgements

This research was supported by National Institute of Health (NIH) grant 1R21NS08064901A1.

References

1. Thomas MA, Yue K, Binesh N, et al. Localized two-dimensional shift correlated MR spectroscopy of human brain. Magn Reson Med. 2001;46(1):58-67.

2. Lipnick S, Verma G, Ramadan S, et al. Echo planar correlated spectroscopic imaging: implementation and pilot evaluation in human calf in vivo. Magn Reson Med. 2010;64(4):947-56.

3. Furuyama JK, Nagarajan R, Roberts CK, et al. A pilot validation of multi-echo based echo-planar correlated spectroscopic imaging in human calf muscles. NMR Biomed. 2014;27(10):1176-83.

4. Lustig M, Donoho D, Pauly JM. Sparse MRI: the application of compressed sensing for rapid MR imaging. Magn Reson Med 2007;58(6):1182–95.

5. Wilson NE, Burns BL, Iqbal Z, Thomas MA. Correlated spectroscopic imaging of calf muscle in three spatial dimensions using group sparse reconstruction of undersampled single and multichannel data. Magn Reson Med. 2015;74(5):1199-208.

6. Goldstein T, Osher S. The Split Bregman method for L1-regularized problems. Siam Journal on Imaging Sciences 2009;2:323-43.

Figures

Figure 1: Schematic diagram showing the accelerated 5D ME-EP-COSI sequence.

Figure 2: The non-uniform sampling mask used for the 4x along the kz-t1 dimensions is shown.

Figure 3: Results from the fully sampled (A), NUS (B), and reconstructed (C) corn oil phantom scans are shown. The olefinic fat (UFD), triglyceryl fat (TGFR), unsaturated fatty acid cross peaks (right - UFR, left - UFL) are labeled.

Figure 4: A) The spatial diagram based off of the methyl fat peaks is shown. B) Extracted 2D COSY spectrum from the soleus muscle. C) Extracted 2D COSY spectrum from the marrow.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
3171