Quiet EPI (QuEPI) for single-shot spin and gradient echo EPI sequences for efficient fetal imaging
Jana Maria Hutter1, Anthony N Price1, Lucilio Cordero Grande1, Emer Judith Hughes1, Kelly Pegoretti1, Andreia Oliveira Gaspar1, Laura McCabe1, Mary Rutherford1, and Joseph V Hajnal1

1Centre for the developing brain, King's College London, London, United Kingdom


Quiet sequences are of particular importance for fetal EPI based imaging, where the necessary protection of the unborn infant can often compromise the efficiency and achievable resolution of the EPI acquisition. This is of particular relevance for connectome type studies, where long functional and diffusion weighted sequences need to be acquired in an efficient and safe way.This abstract presents a quiet SE and GE EPI framework with sinusoidal read-out constant phase and merged crusher strategy, completely flexible and adaptable to the scanner impulse response function leading to a decrease of up to 9dB(A).


While acoustic noise reduction is a general aim for MRI examinations to enhance patient comfort or to avoid disturbance in functional MRI (fMRI) studies, it is of particular importance for successful fetal examinations. The vulnerability of the unborn infant to excessive acoustic noise, leading in the extreme cases (so far not reported in MRI) to hearing loss and shortened gestation [L], puts a particular emphasis on adequate protection levels. As external noise protection with ear-plugs/headphones is not available in-utero, safety of this vulnerable subject group can only be guaranteed by reducing the acoustic noise output of the scanner to increase the natural protection provided by maternal habitus. Single-shot echo-planar-imaging (ssEPI) sequences, which are widely used for fMRI and diffusion MRI (dMRI), can be extremely  noisy and may cause PNS in the mother. Noise reduction tends to prolong the read-out, potentially limiting resolution as well as increasing distortion. Building on previous work [S,Sm,He] we developed a ssEPI acquisition platform both for GE and SE, called QuEPI, which allows a significant noise reduction by reshaping all gradient waveforms and tuning these to the scanner acoustic response function, while keeping the acquisition efficient. It is implemented for all ssEPI based sequences, but here we have focused on fetal dMRI and fMRI studies.


The acoustic noise output for axis $$i=\in \{x,y,z\}$$ is calculated [E] as the convolution of the gradient impulse response function $$$IRF_i(t)$$$ (fixed property of the hardware) and the gradient waveform $$g_i(t): r_i(t)=IRF_i(t)*g_i(t),$$ equivalent to $$R_i(f)=FT\{IRF\}_i(f)*FT\{g\}_i(t).$$ The greatest noise reduction is achieved by modifying the frequency content of the gradient waveforms on all three axis with respect to the IRF, in particular exploiting local minima. Therefore, the read-out was modified from conventional trapezoids to sinusoids [S] with controlled amplitude ramp up/down (Fig.1, bottom). This results in a single dominant frequency, which can be optimized to coincide with a low IRF value (Fig.2a, red arrow). The phase-encoding (PE) blips, contributing to frequency content at double the EPI frequency, were modified to a constant low-amplitude gradient (Fig.2a, arrow), shifted to keep the k-space centres in the read-direction at the same PE locations. The slice-refocusing gradient was modified to a single half-sinusoid with a period that can be matched to a multiple of the read-out period. Finally, specifically for dMRI, acoustic contributions arise from the butterfly crushers around the refocusing pulse. These were combined with the slice excitation-refocusing gradient and stretched out in parallel to the diffusion encoding gradients. The reduction of the available maximal strength for diffusion encoding was minimal, and in part compensated by the freed up time (Fig2a, arrow). QuEPI was implemented together with the required modified gridding strategy on a Phillips Archieva 3T scanner, including in-house implemented Multiband [P], and modified diffusion acquisitions [H]. Fetal GE and SE acquisitions with a FOV of 320x320mm, transverse were acquired on 10 fetuses, GA 24+0-34+2 weeks. Further parameters for dSE-QuEPI include resolution=2.5mm3, partial Fourier=0.87, no SENSE, TE=118ms, TR=2000s, frequency 500 Hz and GE-QuEPI include resolution=2.7mm3, no partial Fourier, no SENSE, TE=59ms, TR=1000ms, TE=50ms. QuEPI Frequency=550 Hz (higher, as the general noise output was lower in GE). The dSE-QuEPI protocol was tested with different variants to evaluate the noise reduction of each QuEPI element: the sinusoidal read-out, constant PE humped re/pre-winders with (i) combined and (ii) standard crushers, the trapezoid-cartesian acquisition at (iii) the same frequency and (iv) as defined with low PNS, gradient slew restriction to 120msT/m for fetal use.

Results and Discussion

The results of the sound measurements for dSE experiments (Table 1), reveal a reduction of 9dB(A) compared to the standard trapezoidal-cartesian settings. If compared to a trapezoidal-cartesian acquisition at the same frequency (500Hz), noise levels were reduced by 5.3dB(A) for the same scan time. The non-compromised image quality for QuEPI and EPI with matched frequency is shown in Fig.3, acquired fetal volumes from a dSE-QuEPI scan and a GE-QuEPI scan in Fig.4. While previous methods are either not available for EPI, influence the applied acceleration factors or increase acquisition time, the proposed QuEPI framework is available for all ssEPI based scans, fully compatible with further acceleration techniques such as partial Fourier, SENSE and multiband. The QuEPI approach provides a flexible platform for comprehensive fetal connectome examinations where high data rate with acceptable acoustic performance is needed as well as other examinations where reduced sound is important. Furthermore, it provides flexibility to optimize and synergistically tune the read-out frequency, diffusion gradients and multiband-blips to the scanner specific transfer function depending on the target acoustic output to achieve optimal combinations of acoustic performance and efficiency.


The authors acknowledge funding from the MRC strategic funds, GSTT BRC and the ERC funded dHCP.


[S] Schmitter, MAGMA21:317, 2008:21:31 [E] Edelstein, MRI20(2):155, 2002 [L] Lecanuet, DevPsychoBiol33(3):203, 1998 [O] Ott, MAGMA,2015 [H]Hutter, ISMRM-SMS-Workshop 2015 [P] Price, ISMRM-SMS-Workshop 2015 [Sm] Smink, ISMRM 2007, 1088 [He] Heismann, MRM73(3):1104, 2015 [Pi] Pierre, ISMRM 2013, 256 [He] Hennel, MRM42(6):10, 1999


Fig.1: Sequence diagrams for GE-EPI and dSE-EPI illustrating the sinusoidal read-out with a single frequency, the ramp-up and ramp-down phases, constant phase gradient and extended spoilers as well as merged crusher strategy for dSE-QuEPI

Fig. 2: Simulations showing a) magnitude spectra of the real scanner waveforms and manufacturer transfer functions (IRF) for dSE-EPI and dSE-QuEPI overlaid with a schematic representation of the A-weighting and b) the calculated acoustic response for all three gradient axis for a transverse scan with Phase-Encoding AP (Read:y, Phase:x, Slice:z)

Fig. 3: Adult data for case i) dSE-QuePI and iii) dSE-EPI sampled at the same read-out frequency of 500 Hz but with respective measured noise levels of i) 101.3 dB(A) and iii) 106.6 dB(A).

Fig. 4: Three-plane results from a fetal scan using dSE-QuEPI with a b-value of b=0 and b=100 (upper row), Singleband-GE-QuEPI and Multiband-GE-QuEPI (lower row)

Table 1: A- weighted acoustic output measurement results for different variants of dSE-QuEPI and dSE-EPI performed using an RF compatible fiber optic microphone (OptiSLM, OptoAcoustics, Israel) in the isocenter of the magnet.

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