0769

Double Echo Steady State (DESS) Cones for Motion-Robust Steady-State Diffusion-Weighted Imaging
Catherine J Moran1, Joseph Y Cheng1, Christopher M Sandino2, Marcus T Alley1, Bruce L Daniel1, Michael Carl3, and Brian A Hargreaves1

1Radiology, Stanford University, Stanford, CA, United States, 2Electrical Engineering, Stanford University, Stanford, CA, United States, 3Global MR Applications and Workflow, GE Healthcare, San Diego, CA, United States

Synopsis

Diffusion-Weighted Imaging (DWI) has shown potential for oncologic imaging without a contrast injection. However conventional methods are limited by low resolution and distortion. Steady-state DWI methods provide 3D, distortion-free images but are susceptible to motion artifacts. A Double Echo Steady State (DESS) method with a conical k-space trajectory is presented and assessed for motion artifact and diffusion-weighting with a focus on breast cancer imaging.

Introduction

Non-contrast-enhanced acquisitions continue to garner interest for oncologic body applications but EPI-based Diffusion-Weighted Imaging (DWI) is limited by susceptibility to motion and off-resonance. Steady-state DWI provides a 3D, distortion-free alternative but is also susceptible to to motion and may not achieve the same degree of diffusion-weighting for certain applications [1,2,3]. In this work we combine Double Echo Steady State (DESS) with a cones k-space trajectory with the goal of achieving higher diffusion-weighting with reduced motion susceptibility in a steady-state sequence. We focus on an initial application in breast MRI where robust non-contrast enhanced imaging could have immediate, wide clinical impact.

Methods

B-value Equivalent DESS Gradient Area:In steady-state DWI methods the relationship between applied diffusion gradient and degree of diffusion-weighting is not as direct as in EPI-DWI. However, on an application-by application basis, with known tissue T1s and T2s, the DESS gradient area required to achieve an equivalent contrast to a specific b-value can be calculated. With tissue values for a malignant lesion (T1 1400 ms, T2 50 ms, ADC 0.8 x 10-3mm2/s) and surrounding fibroglandular tissue (sameT1/T2, ADC 1.5 x 10-3mm2/s) the EPI and DESS tissue signals for a range of b-values and DESS gradient areas were simulated. Simulation results determined the range of DESS diffusion gradient areas in phantom and in-vivo experiments.

Sequence: A DESS acquisition with water-only excitation was implemented with a conical k-space trajectory [4] (Figure 1). Within each TR, the first echo samples a center-out conical interleaf with the second echo samples the same interleaf back in. The unbalanced gradient that imparts diffusion-weighting is applied between the two echoes. For this initial investigation, the diffusion gradient was applied only along the slice select axis and the ordering of the cones was sequential with regridding of the data was performed in BART [5].

Phantom experiments: Signal was measured in a diffusion phantom (model 128, High Precision Devices) at different diffusion gradient areas with DESS cones in comparison to different b-values with EPI-DWI. The acquisitions were performed with a 8-channel head coil (GE Healthcare, Waukesha, WI) at 3T, (MR 750 GE Healthcare, Waukesha, WI) with the following imaging parameters: EPI-DWI: single shot, 128 x 128, 3 mm slices, 20 cm FOV, b-values 200 s/mm2, 600 s/mm2, TR 4000 ms; DESS-Cones: 256 x 256, 3 mm slices, 36 cm FOV, diffusion gradient areas 63 mT/m×s, 188 mT/m×s, FA 15, TR 10 ms. Percent signal difference between high and low diffusion-weighting was calculated in three vials representing a range of Apparent Diffusion Coefficients (ADCs) and compared between the sequences.

In vivo experiments: DESS-Cones, DESS Cartesian, and EPI-DWI acquisitions were performed in an asymptomatic breast volunteer with imaging parameters: DESS 256 x 256, 36 cm FOV, 64-3mm slices, FA 15, diffusion gradient areas 63 mT/m×s, 125 mT/m×s, 188 mT/m×s, EPI-DWI 256 x 128, 36 cm FOV, 42 5 mm slices, b-values 200 s/mm2, 600 s/mm2, 800 s/mm2. Image quality and level of motion artifact were visually assessed between DESS-Cones and DESS-Cartesian acquisitions. Signal level in two ROIs of fibroglandular tissue were measured in both the DESS and EPI-DWI acquisitions. The signal differences in the fibroglandular tissue with increased diffusion-weighting were compared between the sequences and with the simulation predictions.


Results and Discussion

Simulations predicted a diffusion gradient area of 200 mT/m×s with DESS would induce the same degree of contrast (~ 30% signal difference), between a malignant lesion and fibroglandular tissue as a b-value 600 (Figure 2). In the phantom, the relative percent signal difference for different ADCs was similar between EPI-DWI and DESS-Cones. However, the percent signal difference was higher for all three vials with DESS-cones (Figure 3). In vivo, motion artifacts that are evident in the Cartesian DESS acquisitions at all levels of diffusion-weighting and are not present in any of the corresponding DESS-cones images (Figure 4). The most strongly diffusion-weighted images demonstrate low SNR which may improve with further protocol optimization. The signal levels with varying b-values in the EPI-DWI acquisition (Figure 5) reflect those predicted in the simulations. Though the consistency of the signal is better between the ROIs with DESS-cones possibly owing to the elimination of ghosting artifacts both DESS acquisitions indicate lower signal levels than are predicted by the simulations.

Conclusion

DESS-Cones shows potential for reducing ghosting due to motion for a range of diffusion gradient areas, including those necessary to achieve clinically appropriate diffusion contrast in the breast however, excess signal loss in comparison to EPI-DWI must be further characterized.

Acknowledgements

We acknowledge support from GE Healthcare, P41 EB015891 and R01 EB009055

References

1. McNab et al., NMR Biomed. 2010; 23: 781-793.

2. Bieri,et al., Magnetic Resonance in Medicine 2012; 68(3): 720-9.

3. Granlund, et al., Magnetic Resonance Imaging 2014; 32: 330-341.

4. Gurney, et al., Magnetic Resonance in Medicine 2006; 55: 575-582.

5. Uecker, et al., Berkeley Advanced Reconstruction Toolbox, ISMRM 2015, Toronto, Canada.

Figures

Figure 1. DESS acquisition (a) implemented with a conical k-space trajectory (b). For this initial investigation cones interleaves are acquired sequentially.

Figure 2. Simulations of the signal decay for a malignant lesion (ADC 0.8 x 10-3 mm2/s) and fibroglandular tissue (ADC 1.5 x 10-3 mm2/s) in conventional DWI (a) and DESS (b). For DESS gradient area per TR of 200 mT/m⋅sthe predicted percent difference (31%) between the tissues is similar to the predicted percent difference (34%) in conventional DWI for a 600 b-value. Tissues mislabeled in DESS plot - malignant blue, fibrous green.

Figure 3. EPI-DWI (top row) and DESS-Cones (bottom row) in a diffusion phantom at two levels of diffusion-weighting. Gradient areas for DESS-Cones were chosen based on simulations of breast tissue, though T1 and T2 may be different than in tissue. Percent difference of signal between high and low diffusion-weighted images for each acquisition,in 3 of the vials, demonstrate relative signal change reflective of ADC with signal change with DESS cones over estimating degree of diffusion-weighting, assuming EPI-DWI is ground truth. ADC values of vials at room temperature listed at top right.

Figure 4. Motion artifacts (white arrows) are present for a range of diffusion gradient areas in DESS images acquired with a Cartesian k-space trajectory (left column) in a normal breast volunteer. In the corresponding DESS cones images (right column) no motion artifacts are present. Due to the oversampling of central k-space and slightly variable TRs ( 8 msto 13 ms), scan times did vary between the acquisitions (Cartesian DESS scan times: 2:46, 3:00, 3:32, Cones DESS scan times: 3:08, 4:13, 4:13).

Figure 5. Left column shows ROI placement for each acquisition. The measured EPI-DWI signal level at the three b-values (a) reflects that predicted by the simulations (Figure 1a). The signal levels are also consistent between the 2 ROIs. In both DESS Cones (b – solid lines) and DESS Cartesian (b – dotted lines) the signal decrease with increased gradient area is larger than that predicted by the simulations (Figure 1b.), most notably in DESS cones where the decrease is twice as large as that predicted. The measured signal levels also vary between the ROIs, particularly for the DESS Cartesian images.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)
0769