Introduction

Magnetic resonance spectroscopic imaging (MRSI) is a value added modality, augmenting structural scans with metabolic spatial information. However, it usually requires long scan durations. One of the powerful techniques for faster acquisition is so called proton echo planar spectroscopic imaging (EPSI), which delivers spectroscopic and 1D spatial information in a single excitation (Fig. 1a), with additional spatial dimensions undergoing phase encoding^1-3. EPSI trajectories suffer from undesired inconsistencies between readout lines acquired with alternating positive and negative gradients⁴. To overcome this, the positive and negative readouts are conventionally processed individually⁵ and then combined, at the cost of halving the spectral width (SW); or a flyback readout is used⁶, foregoing the negative readouts. An alternative method⁷, which we here term COKE (COherent K-t space EPSI), relies on interleaving “blipped” PE gradients in between readout lines in order to produce coherent phase between the k-t space readout lines (for a given PE). COKE introduces phase inconsistencies along the PE dimension, but these are straightforward to correct as shown in many EPI reconstruction schemes⁸. The full COKE readout (Fig. 1b) doubles SW in comparison to EPSI. In an alternative realization, denoted here as “diet” COKE (Fig. 1c), the SW is halved, yielding the equivalent EPSI dataset at half the acquisition time. In this work, we integrated multi-band CAIPIRINHA slice selection⁹ with the COKE interleaved PE gradients. By introducing dual-band CAIPI COKE (Fig. 1d) and dual-band CAIPI “diet” COKE (Fig. 1e) sequences, one can further accelerate the measurement two-fold.

Methods

We implemented and compared EPSI, COKE, diet COKE, and dual-band CAIPI variants (Figs. 1a-e) in phantoms and in human volunteers at 3T (Tim Trio, Siemens, Erlangen). Note that, in order to integrate the multi-band CAIPIRINHA into COKE, one must switch the slice-selection gradient’s polarity between the even and odd PE lines, respectively, since COKE interleaves the PE lines during each excitation. The reconstruction of the data includes reordering and phase corrections due to PE and CAIPI phase shifts. We compared EPSI, diet-COKE and dual-band diet-COKE in phantoms mimicking brain metabolites and in a fat/water phantom (based on white cheese and oil). The single-band excitation pulse used was an SLR-90° with bandwidth-time product of 18 and duration of 3.8 ms, whereas the dual-band excitation pulse in Fig. 1d,e was based on SLR-90° with a bandwidth-time product of 12 and duration of 5 ms. The metabolite phantom included 10mM L-Glutamic acid, 10mM Creatine 8mM myo-Inositol, 2mM GABA, 2mM Choline chloride, 5mM Choline chloride, 5mM Sodium lactate 12.5mM NAA and 0.5mL Gd for short T₁ (T₁ of water was 625 ms). The relevant scan parameters are summarized in the relevant figure captions.

Results

Fig. 2 shows phantom results comparing “diet” COKE and EPSI. EPSI data was reconstructed using odd readouts only for spectrum quality comparison of COKE reconstruction and equivalent SNR evaluation (having the same as “diet” COKE SNR∕√2). The measured SNR ratio was 1.06, which is close to the expected factor of unity. Fig. 3 demonstrates dual-band “diet” COKE phantom results for spectroscopic imaging, with the brain mimicking phantom and the fat/water phantom. A 5 cm gap between the slices was introduced to ensure their contents differ sufficiently and for clarity of presentation; however, the sequence works also well for adjacent slices (not shown), as was shown before with CAIPIRINHA implementation8. Fig. 4 shows in-vivo brain MRSI acquisition using dual-band COKE implementation, showing it is capable of producing multi-slice spectroscopic images (with an acceleration factor of two due to dual-band and using the double SW versus EPSI).

Conclusions

In this study, we designed a sequence that integrates a multi-band CAIPIRINHA with COKE acquisition for faster MRSI. This was demonstrated in both phantoms and in-vivo. Alternatively, dual-band COKE can be used to double the SW of EPSI, which is a limiting factor in-vivo, particularly at ultrahigh (≥7T) fields where SW requirements can become prohibitive.

Acknowledgements

We are grateful to Osnat Volovyk for preparation of the phantom mimicking brain metabolites. We acknowledge the Monroy-Marks Career Development Fund and the historic generosity of the Harold Perlman Family.

References

[1] Mansfield, P. (1984) Magn. Reson. Med., 1(3), 370-386, [2] Posse, S., et.al (1997) Magn. Reson. Med., 37(6), 858-865, [3] Lin F. H., et.al. (2007) Magn. Reson. Med., 57(2), 249-257, [4] Posse, et.al. (2013) J. Magn. Reson. Imag., 37(6), 1301-1325, [5] Posse S., et al. (1994) Radiology, 192, 733-738, [6] Cunningham C.H. et. al. (2005) Magn. Reson. Med.,54,1286–1289, [7] Webb P., et.al. (1989) Magn. Reson. Med., 12(3), 306-315,[8] Chen N., et. al. (2004) Magn. Reson. Med., 51,1247–1253, [9] Setsompop K., et.al. (2012) Magn. Reson. Med., 67(5), 1210-1224.