Quantification of liver and patient shape changes with abdominal compression on the MR-Linac

Purpose:
Abdominal compression (AC) is used to reduce respiratory motion in patients undergoing abdominal stereotactic radiotherapy. Resultant anatomical changes, including displacements in the direction of organs at risk (OARs) may impact treatment dosimetry. The relationship between the liver and abdominal wall has not previously been examined. This work quantifies changes in liver volume, patient contour, and distance between the anterior liver to the abdominal wall as a result of AC on images acquired on an MR Linac.
Material & Methods:
Eight participants (seven patients, one healthy volunteer) were imaged on a 1.5T MR Linac with and without an AC belt. Images were T2-weighted, in free-breathing (FB), without dietary preparation. Four participants had repeat imaging at a different timepoint. All images were segmented and evaluated by an experienced therapeutic radiographer (RTT) and reviewed by another. Overlap of liver volumes (percentage overlap of structures relative to the union of both structures) with and without AC were calculated for all sessions. Distance between the anterior surface of the liver and the anterior abdominal wall was measured at 2cm intervals from the liver dome inferiorly along the calculated centre of the liver until the liver no longer maintained integrity. For this study, the external volume was delineated between vertebral levels T10 and L2 for equivocal comparison. Variation in patient contour volume was quantified with and without AC, and at different time points.

Results:
24 MRI scans (12 FB, 12 with AC) were acquired for eight participants. There was no statistically significant difference in liver volume (p=0.9) or overlapping volumes for FB or AC (p=0.7). Overlapping volumes are shown in Fig. 1.
For patients undergoing repeat imaging (median gap between sessions = 5 days) there was no statistically significant change in external volume between sessions 1 and 2 for FB (mean decrease 228.3 cm3, SD 184.7, range 62.3-484.0, p=0.09) or with AC (mean 105.8 cm3, SD 159.8, range -73.8-308, p=0.28). The negative value represents an increase, as 1 patient appeared to be compressed less at AC2 than AC1.
The largest variations in liver-abdomen distance occurred >6cm inferior to liver dome, For example, mean change at 0cm was 0.1cm (SD 0.9, range -1.1-1.57) and 0.7 (SD 1.56, range -0.2-4.8) at 6cm (negative value = increase). The increase in liver-abdomen distance was not significant between session 1 and 2 for both FB (p = 0.2-0.9) and AC (p = 0.1-0.9) at each interval.
Conclusion.
This study demonstrates that based on liver and external contours, AC is more reproducible in superior liver lesions. However, variations in belt position and inflation level may lead to anatomical variations for inferior lesions. These results will be validated in a larger cohort as part of ongoing work.