Synopsis

Keywords: Pulse Sequence Design, Lung, UTE, Ultra-Short TE

δUTE sequence collects multiple closely spaced short TE (<2ms) echoes resulting in a more accurate T₂^* mapping of anatomies with short T₂ time. Application to lungs reveals inhomogeneities in T₂^* maps in post-COVID-19 adults.

Introduction

Unlike conventional MRI sequences, Ultra-short TE (UTE) imaging with an echo time of less than 0.1ms allows capturing signals from the anatomies with short $$$T_{2}^{*}$$$ in lungs. Multi-echo UTE sequence produces a set of images with different echo times required for $$$T_{2}^{*}$$$ mapping. However, shorter echo intervals in the multi-echo UTE are restricted due to the $$$dB/dt$$$ hardware limitations. To acquire these shorter TEs by shifting the first echo multiple times requires multiple separate acquisitions and takes a long scan time. To overcome this constraint, we’ve implemented δUTE sequence which allows for closely spaced short TE (<2ms) echoes. The sequence was validated on a phantom with the known $$$T_{2}^{*}$$$ values and applied to study the impact of COVID-19 on adults’ lungs.

Methods

The δUTE sequence was implemented on a clinical 3T scanner (Vantage Galan, version 7, Canon Medical, Japan). The sequence varies the position of the first echo in short increments for the consecutive segments. A phantom study was carried out on a $$$T_{2}^{*}$$$ phantom (Calimetrix, Model 450, USA) with five short (4.44/3.62/2.27/1.89/1.05ms<5ms) inserts. The scanning protocol for the phantom study included: (i) a multi-echo UTE with six TEs=0.096/2.3/4.5/6.7/8.9/11.1ms, TR=14.7ms, NEX=1, FA=4°, FOV=25×25cm in the axial orientation and matrix size 256×256; additional echoes were acquired with (ii) δUTE with four TEs=0.34/0.74/1.14/1.54ms (δTE=0.4ms.). For both series, about eight thousand UTE lines were acquired, and the total scan time was 9 minutes. Two time-dependent signal curves (6 echoes from (i) only and 10 echoes from (i) and (ii)) were fitted into equation [Eq. 1] using the least squares method:
$$S(t)=S(0)\cdot {\exp{(-t/\mathrm{T_2^*)}}} + S' \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad [Eq. 1]$$
The human study included two healthy (25 ± 3 years) and two post-COVID (31 ± 4 years) volunteers, scanned with informed consent. The scanning protocol was similar to the one used for the phantom with the addition of fat-suppression (five SPectral Adiabatic Inversion Recovery (SPAIR) pulses per 48 segments, ~1 pulse per 34 UTE lines). An extra series of 3D UTE without fat suppression (TE/TR=0.096/3.7ms, NEX=1, FA=5°) for an automated lung segmentation was acquired [1]. All series were collected using respiratory gating during expiration with a FOV=50×50cm in the coronal orientation. A time-dependent signal curve of 10 echoes was fitted using a mono-exponential model [Eq. 1] and a bi-exponential model [Eq. 2]:
$$S(t)=S_{s}(0)\cdot {\exp{(-t/\mathrm{T_{2s}^*)}}} +S_{l}(0)\cdot {\exp{(-t/\mathrm{T_{2l}^*)}}} + S' \quad \quad \quad [Eq. 2]$$
which provides more information on the fractions of tightly bound and unbounded water molecules by separating the signal into short $$$T_{2s}^{*}$$$ and long $$$T_{2l}^{*}$$$ components [2].

Results

Schematic pulse sequence diagrams of multi-echo UTE and δUTE are shown in Figures 1a and 1b, respectively. The position of the first echo in the δUTE sequence can be varied for the different segments within a single acquisition. In the diagram, the first echo is shown shifted by TE and possible n^th shift $$$n·δTE$$$ is shown with dashed gradients and thinner line echo. Axial image of $$$T_{2}^{*}$$$ phantom with overlayed colored ROIs corresponding to the five short $$$T_{2}^{*}$$$ inserts is shown in Figure 2a. Figures 2b-k demonstrate all 10 echoes collected with multi-echo UTE (b, g-k) and δUTE (c-f). Note significant drops in signal intensities for ROIs 3 – 5 between TE=0.096 (b) and TE=2.3ms (g). These are consecutive first and second echoes of multi-echo UTE. δUTE echoes (c-f) in contrast provide a more gradual decrease in signal intensities allowing to estimate relaxation times with higher precision. Comparisons between six-echo and ten-echo fits for five phantom inserts are shown in Figure 3a-e. Six-echo fits are given in blue while ten-echo fits are given in red with δUTE echoes denoted with red squares. Discrepancies in the fits are best noticeable for inserts with $$$T_{2}^{*}$$$ of 2.27 and 1.89 ms. Figure 3f shows the plot of residuals between true and estimated $$$T_{2}^{*}$$$ values. Results of the human study are shown in Figure 4 colormaps: the top row for healthy and the bottom row for post covid subjects. The first column shows a single slice $$$T_{2}^{*}$$$ colormap for mono-exponential fit, bi-exponential fit colormaps with short and long $$$T_{2s,l}^{*}$$$ are shown in the second and third columns while the last column shows the fraction of $$$T_{2s}^{*}$$$. All the colormaps of healthy volunteers were more homogeneous compared to those of post-COVIDs.

Discussion

Additional echoes close to the shortest of the UTE echoes result in a more accurate estimation of $$$T_{2}^{*}$$$ maps for tissues with short $$$T_{2}^{*}$$$. UTE images with high temporal echo time resolution were previously collected ex-vivo for various specimens and cadaveric tissues in tendon and knee [3] as well as in rodent lungs under anesthesia [4]. The current study shows the feasibility of $$$T_{2}^{*}$$$ mapping using δUTE imaging with high temporal echo time resolution on in-vivo human lungs.

Conclusion

The δUTE sequence was implemented and tested on the short $$$T_{2}^{*}$$$ phantom. The sequence was successfully applied for lung imaging; the bi-exponential signal model fit may provide valuable insides into the COVID impact on human lungs in future studies.

Acknowledgements

This work was supported by Canon Medical Systems, Japan (grant 35938).

References

[1] Malis V., et al., in ISMRM (2021).
[2] Biswas R., et al., Bone. 50(3), 749-55. (2012).
[3] Carl M, et al., Magn Reson Med. 76(2), 577-82 (2016).
[4] Takahashi M, et al., J Magn Reson Imaging. 32(2), 326-33 (2010).