Purpose

Geometric accuracy is one major concern when applying MRI for radiotherapy purpose. MR geometric distortion is related to various factors, and B₀ inhomogeneity (ΔB₀) is one of them. Besides ΔB₀ due to magnet imperfection, B₀ inhomogeneity can be also related to organ shape, tissue-specific magnetic susceptibility and respiration. Recently, change in the ΔB₀ due to respiratory states was reported in the breast [1]. Dynamic geometric distortion in the lung has been investigated via ΔB₀ simulation in another study [2]. However, in vivo ΔB₀ and the associated geometric distortion in the liver have rarely been reported. In this study, we aimed to evaluate the geometric distortion of liver in vivo due to ΔB₀ variation at inhalation and exhalation breath-hold.

Method

Liver imaging was conducted in 7 healthy volunteers using a 1.5T Siemens MR scanner (Aera, Siemens Healthineers, Erlangen, Germany). Coronal images at inhalation breath-hold and exhalation breath-hold were acquired using a dual echo GE sequence (TR=70-137ms, TE1/TE2 = 4.76/9.52ms, slice thickness = 8-10mm, 5-11 slices, 38-48cm FOV, matrix=192x192 or 110x110, Bandwidth=260 or 1000Hz/pixel, phase encoding (PE) direction = left-right). The corresponding ΔB₀ map was first calculated from phase images using customized Matlab script (Mathworks, Natick, MA) and then geometric distortion along frequency encoding (FE) direction (Δd_FE) was derived. For each image set, three square ROIs, with two ROIs at the right liver lobe (proximal and distal to the lung, hereafter RLp and RLd) and one ROI at the left liver lobe (LL), were drawn (Figure 1). To avoid phase wrapping, all ROIs were chosen away from the interface with large susceptibility difference. A Wilcoxon signed-rank test was performed to compare the ΔdFE at inhalation and exhalation breath-hold in three ROIs.

Results

As illustrated in Figure 2, the measured Δd_FE at inhalation in the RLd was significantly higher than that at exhalation (0.21±0.16mm and 0.13±0.08mm, P<0.05). Larger Δd_FE was also observed in the RLp (inhalation: 0.64±0.44mm and exhalation: 0.53±0.39mm, P=0.11) and LL (inhalation: 0.16±0.11mm and exhalation: 0.13±0.16mm, P=0.30). Significantly higher Δd_FE was also observed in RLp (inhalation: 0.64±0.44mm and exhalation: 0.53±0.39mm) when comparing to LL (inhalation: 0.16±0.11mm, P<0.05 and exhalation: 0.13±0.16mm, P<0.05) and RLd (inhalation: 0.21±0.16mm, P<0.05 and exhalation: 0.13±0.08mm, P<0.05). When comparing between RLd and LL, no significant difference in Δd_FE was observed at inhalation (RLd: 0.21±0.16mm and LL: 0.16±0.11mm, P=0.14) and at exhalation (RLd: 0.13±0.08mm and LL: 0.13±0.16mm, P > 0.05).

Discussion

In this study, we have presented the geometric distortion pattern in the liver due to ΔB₀ variation at inhalation and exhalation. For all ROIs, larger Δd_FE was observed at inhalation breath-hold than at exhalation breath-hold, which may be related to the amount of air presence in the lung. As illustrated in Figure 3, the largest geometric distortion was observed mostly at the location with large magnetic susceptibility difference (e.g. liver-lung interface). The largest Δd_FE was observed at the RLp because RLp was chosen at a location close to the liver-lung interface. On the other hand, Δd_FE of LL was smaller than Δd_FE of RLp though ROIs of LL and RLp were chosen at a similar proximity to the lung. The smaller Δd_FE observed in LL than in RLp might be explained by the lower tissue magnetic susceptibility difference between the heart and the liver. Note that LL is more prone to cardiac motion but this passive motion of LL should be negligible in the ΔTE of 4.76ms. There are several limitations in this study: (1) the mismatch in the each ROI location between inhalation and exhalation breath-hold was ignored; (2) geometric distortion could only be assessed along FE direction and (3) small sample size used in this study. Nonetheless, compared to the reported 12 to 26 mm of the liver respiratory motion along SI by Rohlfing et al [3], our observed Δd_FE was generally small (mostly less than 1mm) for both inhalation and exhalation breath-hold for all ROIs, indicating the small contribution of geometric distortion due to ΔB₀ variation to the positional uncertainty of the liver during respiration.

Conclusion

The geometric distortion in liver due to ΔB₀ variation at inhalation and exhalation has been obtained in this study. For all ROIs and both respiratory states, the calculated Δd_FE (along SI direction) was generally less than 1mm.

Acknowledgements

No acknowledgement found.

References

[1] Bolan PJ, Henry P, Baker EH, et al. Measurement and correction of respiration-induced B0 variations in breast 1H MRS at 4 Tesla. MRM 52: 1239-1245 (2004)

[2] Stanescu T and Jaffray D. Investigation of the 4D composite MR image distortion field associated with tumor motion for MR-guided radiotherapy. Medical Physics 43: 1550-62 (2016)

[3] Rohlfing T, Maurer C, O’Dell W, et al. Modeling liver motion and deformation during the respiratory cycle using intensity-based registration of gated MR images. Med. Phys. 32: 427-32 (2004)