Effect of Gadolinium-Induced Susceptibility on First-Pass Single-echo Dixon CE-MRA and Methods for Correction
Eric G. Stinson1, Joshua D. Trzasko1, and Stephen J. Riederer1

1Radiology, Mayo Clinic, Rochester, MN, United States

Synopsis

Single-echo Dixon imaging for contrast-enhanced MR angiography (CE-MRA) can provide the advantages of multi-echo Dixon without the tradeoff of longer acquisitions and reduced temporal resolution. Single-echo Dixon imaging assumes that both the water and fat signals are real and have known initial phase and phase due to field inhomogeneities. However, when a paramagnetic Gadolinium-based contrast agent is injected for CE-MRA, the field may be perturbed and render the a priori phase estimates invalid. The purpose of this study is to demonstrate the effect of Gd-induced field perturbations on first-pass single-echo Dixon CE-MRA and describe strategies to avoid or correct artifacts.

Purpose

Multi-echo Dixon methods (1) applied to contrast-enhanced MR angiography (CE-MRA) avoid misregistration-based subtraction artifacts (2) and improve signal-to-noise ratio (3) compared to traditional time-subtraction methods. If initial phase $$$\phi_0$$$ and $$$\Delta B_0$$$-induced phase $$$\phi(t)$$$ are known a priori, single-echo Dixon methods can provide the same benefits (3) while offering temporal advantages, particularly in time-resolved studies, where the additional calibration time needed to estimate $$$\phi_0$$$ and $$$\phi(t)$$$ is amortized over numerous time points. However, when gadolinium-based contrast material is used, the paramagnetic Gd3+ ions may perturb the field and render the a priori phase estimates invalid (4). This effect is the basis of dynamic susceptibility contrast MRI (DSC-MRI), but in single-echo Dixon imaging manifests as reduced CE blood signal in the water image as it leaks into the fat image. The purpose of this study is to demonstrate the effect of Gd-induced field perturbations on first-pass single-echo Dixon CE-MRA and describe strategies to avoid or correct artifacts.

Methods

Theory: Single-echo Dixon imaging assumes that both the water (W) and the fat (F) signals are real and have known $$$\phi_0$$$, $$$\phi(t)$$$, and chemical shift-induced phase $$$\theta(t)$$$ (Eq. 1). The phase terms can be found with a multi-echo calibration scan (5,6), or through a "virtual shimming" procedure (7). Constraining W and F to be real and performing a least-square estimate allows reconstruction using Eq. 2 (3,8). However, in the presence of a Gd-based contrast agent, the phase may be perturbed by $$$\delta \phi$$$ (Eq. 3) and reconstruction via Eq. 2 may result in incorrect W and F estimates. Evidence of this field perturbation is shown in Fig. 1, where the estimated background phase within a contrast-enhanced vial of b-gel varies with Gd concentration.

$$G(t) = \left( W + F e^{i\theta(t)} \right) e^{i (\phi(t) + \phi_0)} + N \hspace{1cm} \text{(Eq. 1)}$$

$$\begin{bmatrix}W \\ F \end{bmatrix} = \left[ Re\{ A^{*} A\} \right]^{-1} Re\left\{ A^* G \right\} \hspace{1cm} \text{(Eq. 2a)}$$

$$A=e^{i (\phi(t) + \phi_0)} \begin{bmatrix} 1 & e^{i\theta(t)} \end{bmatrix} \hspace{1cm} \text{(Eq. 2b)}$$

$$G(t) = \left( W + F e^{i\theta(t)} \right) e^{i (\phi(t) + \phi_0 + \delta \phi)} + N \hspace{1cm} \text{(Eq. 3)}$$

Artifact Avoidance: We have identified three strategies for avoiding artifacts from susceptibility artifacts from Gd inflow: (i) Use a lower contrast dose to lessen the field perturbation, (ii) Acquire the multi-echo calibration scan after contrast has reached equilibrium (estimate a semi-perturbed field), (iii) Use a hybrid reconstruction technique (Fig. 2) to virtually shim (7) the field map within the water image mask of each single-echo acquisition.

In Vivo Experiments: An IRB-approved experiment was performed with 10mL and 18mL of contrast material injected on different days with the same volunteer. Unaccelerated coronal thigh images were acquired with a 10 channel phased array at 3.0T (GE, Waukesha, WI) with the following scan parameters: TE1/TE2/TR=2.3/3.5/5.6msec, α=18°, BW=±62.5kHz, FOV=38×38×20cm, Matrix=240×240×100. These multi-echo images were acquired pre- and post-injection of MultiHance (Bracco Diagnostics, Princeton, NJ, USA) to estimate $$$\phi_0$$$ and $$$\phi(t)$$$. A fully-sampled single-echo image was acquired with identical parameters except: TE/TR=2.8/4.9msec. The 18mL single-echo image was reconstructed using phase derived from the pre- and post-contrast multi-echo images and with the hybrid reconstruction (Fig. 2) .

Results

Targeted maximum intensity projections (MIPs) of results are shown in Fig. 3. When 18 mL of contrast is used, signal loss is seen within the femoral artery (Fig 3a, arrowheads), but is partially mitigated by using method (ii) (Fig 3b). Further improvement in luminal signal is realized by method (iii) (Fig. 3c). Using 10 mL of contrast material (method (i)) also prevents Gd-induced signal loss (Fig. 3d), but, results in lower vascular signal (signal profiles, Fig. 3). The signal profiles (Fig. 3, bottom) show that each of strategy prevents or corrects for Gd-induced susceptibility effects.

Discussion

Three strategies for avoiding Gd-induced susceptibility effects in first-pass single-echo Dixon CE-MRA have been shown. Reducing the contrast dose decreases field map perturbations and retains contrast-enhanced blood signal. Phase derived from the post-contrast images is a better approximation of the perturbed phase and mitigates some signal loss, but while not seen here, could potentially introduce errors from Gd enhancement in the veins. The hybrid reconstruction technique further corrects the field map perturbations, but due to the strict masking step may not successfully handle voxels containing both water and fat signal. Future work will explore more robust virtual shimming techniques, other methods of estimating the perturbed field map, and additional uses of the estimated $$$\delta \phi$$$ map.

Conclusion

Gd-induced susceptibility can have a non-negligible effect on first-pass single-echo Dixon CE-MRA, but this can potentially be alleviated using one of the three presented methods.

Acknowledgements

Funded by: NIH EB000212, NIH RR018898

References

1. Dixon, W. Radiology 153, 189–194 (1984).

2. Leiner, T. et al. Eur. Rad. 23, 2228-2235 (2013).

3. Stinson, E. et al. MRM. 74, 81-92 (2015).

4. Hoory, T. et al. MRM. 59, 925–929 (2008).

5. Yu, H. et al. MRM. 55, 413–422 (2006).

6. Ma, J. JMRI. 27, 881–890 (2008).

7. Xiang, Q.-S. Proc. ISMRM 789 (2001).

8. Bydder, M. et al. MRI. 29, 216–221 (2011).

Figures

Figure 1: Plots of the estimated background phase within Gd-enhanced bovine gelatin (solid line) and unenhanced bovine gelatin in a nearby part of the phantom (dashed line). There is a clear trend of additional phase accrual with increasing concentration of Gadolinium. On the other hand, unenhanced gelatin in the same images doesn't show the trend, allowing systematic errors to be ruled out.

Figure 2: Flowchart of the hybrid reconstruction technique used in this work.

Figure 3: Targeted MIPs of 18mL (a-c) and 10mL (d) Gd-enhanced single-echo Dixon studies. Luminal signal smoothness and signal loss is seen in (a) where phase terms were estimated from pre-contrast data. Post-contrast-estimated phase terms (b), return some signal. Virtual shimming (c) further increases signal. Signal loss is also avoided by using 10mL of contrast as in (d). The signal profiles at the bottom of the figure, corresponding to the dotted lines in the images, illustrate this further.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
2673