Purpose: We developed the first hands-free, MR-compatible, and real-time e4D ultrasound probe for application in imaged guided radiotherapy (IGRT). This work assessed the correlation between salient anatomical features in simultaneously acquired and synchronized ultrasound and MR temporal data of the liver with the ultimate goal of implementation in motion management for radiotherapy.
Methods: Simultaneous 2D MR and 3D ultrasound images were acquired from five healthy volunteers. All MR data in this study were acquired on a 3T MRI (Premier, GE Healthcare, Waukesha, WI) using with a fast spoiled gradient echo sequence (FSPGR). Consecutive slices of 2D sagittal MR images (200 frames/slice) spanning the liver were acquired for each volunteer. The ultrasound was positioned such that liver vasculature could be imaged through the entire session. The ultrasound and MR datasets were streamed to a separate machine for synchronization and analysis. For both the MR and ultrasound data, vessels were identified and tracked through valid sequences. The Pearson correlation coefficient between the total displacement of a vessel in ultrasound and a vessel in MR was calculated. The peaks of the motion traces were analyzed and compared between modalities in order to assess respiratory phase shift.
Results: There was variability in the correlation coefficients between volunteers. The highest and lowest mean correlations over all sequences were 0.850 and 0.514, respectively. The median phase shift per sequence varied from 0 to 2 frames. Sequences with low MR/ultrasound correlation tended to have increased phase shift. Additionally, it was observed that sequences with small ranges of motion tended to have lower correlation.
Conclusion: Good correlation between imaging modalities was found when breathing motion was sufficiently regular. Furthermore, minimal phase shift was detected. Therefore, simultaneous ultrasound and MR of the liver were successfully acquired and synchronized making this technique a viable solution to motion management in IGRT.