1896

Feasibility of 3D PREFUL: 3D dynamic lung ventilation imaging, initial comparison to 2D PREFUL in healthy volunteers
Filip Klimeš1,2, Andreas Voskrebenzev1,2, Marcel Gutberlet1,2, Grzegorz Bauman3,4, Oliver Bieri3,4, Robert Grimm5, Tawfik Moher Alsady1,2, Frank Wacker1,2, and Jens Vogel-Claussen1,2

1Institute of Diagnostic and Interventional Radiology, Hannover Medical School, Hannover, Germany, 2Biomedical Research in Endstage and Obstructive Lung Disease Hannover (BREATH), German Centre for Lung Research (DZL), Hannover, Germany, 3Department of Radiology, University of Basel Hospital, Basel, Switzerland, 4Department of Biomedical Engineerings, University of Basel, Basel, Switzerland, 5Siemens Healthcare GmbH, Erlangen, Germany

Synopsis

Pulmonary ventilation assessed by Fourier Decomposition (FD) is limited by its 2D acquisition and does not consider dynamics of ventilation. In this work, a method to assess dynamic lung ventilation in 3D using self-gating and phase-resolved functional lung imaging (PREFUL) is presented. The full respiratory cycle was reconstructed and dynamic regional ventilation (RV) maps were generated. Mean slice correlation and anterior-posterior gradient of ventilation values were compared between the 3D and the 2D PREFUL approaches. 3D PREFUL imaging was able to reconstruct dynamic imaging of the respiratory cycle, generate dynamic RV maps and provide good agreement with the 2D approach.

Introduction

Fourier Decomposition (FD)1 has been demonstrated to deliver quantitative measurement of pulmonary ventilation2,3. Recently, a 2D post-processing approach, phase-resolved functional lung imaging (PREFUL) was introduced in order to increase temporal resolution and gain quantitative regional information about perfusion and ventilation dynamics4. One limitation of PREFUL is its 2D acquisition, which is time demanding if the coverage of the whole lung volume is required. 3D self-navigated approaches were previously developed to decrease measurement times and to allow for quantification of lung ventilation5,6 however, ventilation dynamics was not assessed. The purpose of this work was to demonstrate the feasibility of 3D PREFUL measurement, to quantify dynamics of ventilation and to compare 3D PREFUL to 2D PREFUL.

Methods

Six healthy volunteers (3 female, 3 male, age range: 31-50 years) underwent imaging on a 1.5T MR-scanner (MAGNETOM Avanto, Siemens Healthcare, Erlangen, Germany). Firstly, for 2D PREFUL, 200 images per slice were acquired for eight coronal slices of the lungs using an ultra-fast balanced steady-state free-precession (uf-bSSFP) sequence7 with the following parameters: TE 0.64 ms, TR 1.41 ms, TA 180 ms, TW 138 ms, leading to temporal resolution of 318 ms per image, flip angle 35°, FOV 50 x 50 cm2, slice thickness 15 mm, matrix size 128x128 interpolated to 256x256, pixel bandwith 2055 Hz/pixel, total acquisition time 64 seconds per slice. For the 3D approach, 4000 spokes were acquired using a prototype stack-of-stars spoiled gradient echo sequence with golden-angle increment (FOV 50 x 50 cm2, matrix size 128x128 interpolated to 256x256, slice thicknes 15 mm, TE 1.3 ms, TR 3 ms, flip angle 5°) over a period of 3.2 min. The DC signal was used to divide the spokes over 10 respiratory states8, and the images were reconstructed off-line using iterative reconstruction in Berkeley Advanced Reconstruction Toolbox (BART)9. The image registration using group-oriented approach10 towards an intermediate respiratory state was performed by using Advanced Normalization Tools (ANTs)11. Regional ventilation (RV) maps were computed for each respiratory state N using a correction factor for fractional ventilation (FV) accounting for quantification errors due to registration12: $$$ RV(N) = \frac{S_{Exp} - S_{N}}{S_{Exp}} * \frac{S_{Mid}}{S_{N}}= FV(N)*\frac{S_{Mid}}{S_{N}} $$$, where SMid represents MR signal of the intermediate state, SN is the MR signal of N-th state, and SExp represents MR signal of the end-expiratory state. After quantification of RV dynamics, a semi-automated segmentation was used to exclude large vessels. The Pearson correlation of mean regional 2D / 3D ventilation was assessed. Additionally, physiological plausibility in anterior-posterior (A-P) direction of both methods was tested.

Results

Figure 1 shows morphological images (A / C) of the whole respiratory cycle and corresponding dynamic RV maps (B / D) assessed with 3D / 2D PREFUL approach. Similar visual appearance is seen in Figure 2 between 2D (A) and 3D (B) regional ventilation maps, with higher ventilation values of the 3D approach and more pronounced registration artifacts preferably near diaphragm and large vessels in the 3D method. Good agreement of mean RV of both approaches was observed (R = 0.66, p-value = 0.15, see Figure 3). For the 2D approach, no evident pattern was seen in the evaluation of mean RV values as a function of slice location (Figure 4). In 3D imaging, three out of six healthy volunteers showed increased values of ventilation in the posterior slices, whereas in the 2D approach rather homogenous RV values in the A-P direction were observed.

Discussion

The novel 3D PREFUL technique was able to deliver quantitative information about lung ventilation with good agreement with the 2D PREFUL approach in six healthy volunteers. Increased RV values calculated using 3D PREFUL may be explained by lung motion, which is likely better captured with the 3D technique, and by the selection of spokes/images used for RV quantification. Variability of respiration during the 2D slice scans and partial-volume effects at the most anterior and posterior slices are probably the reason for the lack of a pronounced A-P ventilation gradient in 2D imaging. On the contrary, a smoother trend of ventilation distribution and a physiological A-P ventilation gradient was seen in three healthy volunteers with the 3D approach. However, image artifacts suggest that improvements in 3D registration and reduction of echo time are required to increase the image quality and to achieve a more accurate RV quantification of the 3D PREFUL approach.

Conclusion

The 3D PREFUL approach represents an alternative technique to asses lung ventilation dynamics and may reduce the measurement time to three and a half minutes, which might be beneficial in clinical routine.

Acknowledgements

This work was funded by the German Center for Lung Research (DZL) and supported by Siemens Healthcare GmbH. Furthermore, the authors would like to thank Agilo Kern, Arnd Obert and Lea Behrendt for postprocessing support and useful discussions.

References

1. Bauman G, Puderbach M, Deimling M, et al. Non-contrast-enhanced perfusion and ventilation assessment of the human lung by means of fourier decomposition in proton MRI. Magn Reson Med. 2009;62(3):656-664. doi:10.1002/mrm.22031.

2. Zapke M, Topf H-G, Zenker M, et al. Magnetic resonance lung function – a breakthrough for lung imaging and functional assessment? A phantom study and clinical trial. Respir Res. 2006;7(1):106. doi:10.1186/1465-9921-7-106.

3. Kjørstad Å, Corteville DMR, Henzler T, et al. Quantitative lung ventilation using Fourier decomposition MRI; comparison and initial study. Magn Reson Mater Physics, Biol Med. 2014;27(6):467-476. doi:10.1007/s10334-014-0432-9.

4. Voskrebenzev A, Gutberlet M, Klimeš F, et al. Feasibility of quantitative regional ventilation and perfusion mapping with phase-resolved functional lung (PREFUL) MRI in healthy volunteers and COPD, CTEPH, and CF patients. Magn Reson Med. 2017;00(May):1-9. doi:10.1002/mrm.26893.

5. Voskrebenzev A, Gutberlet M, Wacker F, et al. 3D Lung Ventilation 1H Imaging Using a Respiratory Self-Navigated Stack-of-Stars Sequence in Comparison to 2D Fourier Decomposition. Proc Intl Soc Mag Reson Med 24. 2016:2913.

6. Pereira LM, Weng AM, Wech T, et al. UTE-SENCEFUL: high resolution 3D ventilation weighted maps. Proc Intl Soc Mag Reson Med 26. 2018:2466.

7. Bauman G, Pusterla O, Bieri O. Ultra-fast Steady-State Free Precession Pulse Sequence for Fourier Decomposition Pulmonary MRI. Magn Reson Med. 2016;75(4):1647-1653. doi:10.1002/mrm.25697.

8. Feng L, Axel L, Chandarana H, Block KT, et al. XD-GRASP: Golden-angle radial MRI with reconstruction of extra motion-state dimensions using compressed sensing. Magn Reson Med. 2016;75(2):775-788. doi:10.1002/mrm.25665.

9. Uecker M, Ong F, Tamir J, et al. Berkeley Advanced Reconstruction Toolbox. Proc Intl Soc Mag Reson Med 23. 2015:2486.

10. Voskrebenzev A, Gutberlet M, Kaireit TF, et al. Low-pass imaging of dynamic acquisitions (LIDA) with a group-oriented registration (GOREG) for proton MR imaging of lung ventilation. Magn Reson Med. 2017;78(4):1496-1505. doi:10.1002/mrm.26526.

11. Avants BB, Tustison NJ, Song G, et al. A reproducible evaluation of ANTs similarity metric performance in brain image registration. Neuroimage. 2011;54(3):2033-2044. doi:10.1016/j.neuroimage.2010.09.025.

12. Klimes F, Voskrebenzev A, Gutberlet M, et al. Correction for Ventilation Quantification Errors due to Registration in Pulmonary Lung MRI Fourier Decomposition. Proc Intl Soc Mag Reson Med 26. 2018:4358.

Figures

Figure 1: Morphological images of ten respiratory phases reconstructed with the 3D (first row-A) and 2D PREFUL (third row-C) approach and respective dynamic RV maps of the 3D (second row-B) and 2D (fourth row-D) approach. Lack of contrast due to longer TE and registration problems in 3D approach might cause worse vessel sharpness visible in morphological and RV maps of the 3D approach.

Figure 2: Regional ventilation (RV) maps generated with the newly introduced 3D PREFUL approach (first row-A) and with the 2D PREFUL approach (second row-B). Please note unexpected holes at the bottom of the right lobe in slice 4 and 5 for 3D approach, which might be caused by imperfect registration.

Figure 3: Pearson correlation analysis of mean RV values of 3D (RV3D) and 2D (RV2D) averaged over all slices for each volunteer. Each point represent mean RV of a single volunteer derived by 3D and 2D approach. Increased ventilation values of the 3D approach compared to the 2D approach are observed.

Figure 4: Distribution of RV values in anterior-posterior (A-P) direction for 2D method (A) and 3D method (B). Each line represents single volunteer. Homogenous ventilation distribution in A-P direction is seen for 2D approach and the curves of 3D approach show smoother trend (especially in Volunteer 3 and Volunteer 6) compared to 2D approach.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)
1896