0205

Simultanous-multi-slice and Alternating Multi-echo Measurement Sequence (SAME) for Perfusion Imaging

Elias Kellner¹, Marco Reisert¹, Benedikt A Poser², Irina Mader³, Valerij G Kiselev¹, and Michel Herbst¹

¹Department of Radiology, University Medical Center Freiburg, Freiburg, Germany, ²Department of Cognitive Neuroscience, Maastricht Brain Imaging Centre (MBIC), Maastricht University, ³Department of Neuroradiology, University Medical Center Freiburg, Freiburg, Germany

Synopsis

In this work, we combine simultaneous multi-slice acquisition with a multi-echo readout, dedicated to dynamic susceptibility-contrast perfusion imaging (DSC). With this approach, multiple spin and gradient echo images can be obtained at short repetition times to determine both T2 and T1 effects of contrast agent in a robust and stable manner.

Purpose

Perfusion imaging is widely performed using single-echo EPI sequences. More advanced dual- or multiecho sequences in which both, gradient and spin echoes are acquired, allow for assessment of the vessel size index (1), and to model T1-effects (2) but suffer from too long repetition times to accurately sample the blood dynamics. In this work we use a simultaneous multi slice approach to reduce the repetition time by a factor of three and combine it with a scheme of alternating gradient –and spin echo acquisitions.

Methods

A multi-echo SMS-EPI sequence was modified to apply an 180deg refocusing pulse after an initial gradient echo. This pulse is then alternatingly applied during the runtime of the sequence, leading to 6 different contrasts (Figure 1). Simultaneous multislice (SMS) excitation and refocusing pulses were generated as linear combinations of three sinc and refocusing Shinnar–Le Roux (SLR) pulses respectively, using optimized relative phases to reduce peak RF. Peak RF power of the refocusing pulse was further minimized by reducing the amplitude of the slice-selective gradient using the variable rate selective excitation (VERSE) method. The excitation pulse was calculated online and the refocusing pulse was designed offline in Matlab and imported to the sequence during runtime. Online SMS reconstructions of single-shot data were performed using the sliceGRAPPA implementation as distributed with blipped-CAIPI EPI.We performed measurements in 8 patients with no vascular diseases undergoing post-contrast examinations (0.1mmol/Kg Gadovist), on a SIEMENS Prisma, 3Tesla. Protocol parameters were: TR=600ms, TE1=22ms, TE2=55ms, TE3=84ms, Matrix Size 96x96, SMS Factor 3, 15 Slices, 2 x 2 x 5 mm3, GRAPPA Factor 2. For all 6 echoes, signal time courses were converted to estimated changes in relaxation rates using

$$S(t) = S_0(t) \mathrm{e}^{-dR_2(t) TE}$$ where $$$S(t)$$$ denotes the time series of dynamic scans and TE refers to the echo time within a scan. T1 effects can be approximated with the time-dependent $$$S_0(t)$$$, and fitting both $$$log(S_0(t))$$$ and $$$dR2(t)$$$ for each time point t, using a two-parameter linear fit for all echoes, similar to (2).

Results and Discussion

The results for one patient are exemplarily shown in Figure 2-5. The alternating application of spin-echo refocusing pulses leads to an alternating signal level, with each branch having its own steady state. Treating each branch independently results in a total of 6 curves of relaxation rate changes (Figure 3) at an effective sampling rate of 2TR = 1200ms. Both figures indicate that the method yields stable and robust time series for multiple echoes with a single measurement sequence. Note that in Figure 4, relaxation rates were calculated the "standard" way for each echo individually using a fixed $$$S_0$$$ determined from the baseline signals. This does not account for T1 effects, which becomes obvious in slightly different post-contrast signal levesls The effect is even stronger in a pericranial region outside of the skull, where the blood brain barrier is absent. Figure 5 shows that obviously, T1 and T2 effects can be separated in a robust manner using a time dependent $$$S_0(t)$$$, fitted to all echoes. In this work, we only used the gradient echo data for the fit. A more advanced fit accounting also for the microvascular content of the asymmetric- and spin echoes is part of future work.

Conclusions

The proposed method allows for robust and stable measurement of multiple gradient- asymmetric- and spin echoes. With this, both T2 and T1-effects of contrast agent can be separated at good SNR and short repetition times for accurate perfusion measurement implementing both DSC and the DCE in the same examination.

Acknowledgements

This work has partly been funded by German Research Foundation (DFG) grant number KI1089/3-2 .

References

1. Troprès, Irène, et al. "Imaging the microvessel caliber and density: principles and applications of microvascular MRI." Magnetic resonance in medicine 73.1 (2015): 325-341.

2. Kiselev, Valerij G, Schmiedeskamp, Heiko. Chapter "Vessel Size Imaging" in: Perfusion Imaging: Clinical Applications and Theoretical Principles, Wolters Kluwer Health, Editor: Bammer, R, 2016 https://books.google.de/books?id=fxKtCwAAQBAJ

2. Schmiedeskamp, Heiko, et al. "Combined spin-and gradient-echo perfusion-weighted imaging." Magnetic resonance in medicine 68.1 (2012): 30-40.

Figures

Sequence Diagram: In this multi-echo SMS sequence, the refocusing pulse is applied alternatingly every second scan repetition (TR), resulting in 6 different echoes: 4 gradient echoes, one "mixed" echo, one spin echo.

Three images are acquired at echo times 22, 55 and 84 ms (left to right columns) for each dynamic scan. The alternating application of the 180 degree refocusing pulse results in different contrast for even and odd dynamic scan numbers (red and blue row). Without refocusing pulse, three gradient-recalled echo images are acquired, with application of the spin echo pulse, the first echo remains a gradient echo (GE), the second one represents a “mixed” contrast of spin and gradient echo (ME), and the third one represents a spin echo (SE), resulting in 6 different contrast.

Signal time course in a white matter region. Alternating application of the 180deg refocusing pulse with TR=1200ms leads to an alternating steady state for even and odd scan numbers, for each branch. During the first 8 seconds, the pulse is always applied. After that, the signal magnitude alternates, as the 180deg pulse does not only refocus the spins in the transversal plane, but also has the effect of inverting the longitudinal magnetization. This results in a lower signal amplitude in echoes following the next excitation. However, each branch can be treated individually, resulting in 6 different time series(see Figure 4).

Signal time courses converted to changes in relaxation rates for a white matter, and a pericranial region. All echos are treated as individual time courses for both, even and odd scan numbers, resulting in 6 curves. All gradient echo curves are similar, only GE3 at TE=84ms shows a slightly underestimated peak due to signal voids at high concentrations. The T1 effect is visible in different post-contras levels. This is even more obvious in a pericranial region outside of the skull, where the blood-brain-barrier is missing. T1 effects can be accounted for by fitting the full signal model, see Figure 5.

Figure 5. Fitted $$$dR2^*(t)$$$ and $$$S_0(t)$$$ using the relation $$$S(t) = S_0(t) \mathrm{e}^{-dR_2^*(t) \, TE}$$$, derived from the gradient-echo images. $$$S_0(t)$$$ is arbitrarily scaled. Without leakage, the T1-contribution in a white matter voxel is only subtle; it is bounded by the capillary volume on the order of < 5%. In the pericranial region, the leakage results in a steady increasing $$$S_0(t)$$$.

Proc. Intl. Soc. Mag. Reson. Med. 25 (2017)

0205