Patient Setup Management for Pancreatic Stereotactic Body Radiation Therapy: Interfraction Adjustment Between Spine Alignment and the Pancreatic Fiducial During Breath-Hold kV Imaging

Abstract

Active Breathing Coordinator (ABC) is a widely used motion management strategy for the radiation therapy of mobile tumors. During treatment, aligning the patient based on spine alone does not automatically lead to an acceptable alignment with respect to soft tissue targets and adjustment is necessary. Determining the magnitude of such an adjustment is an important quality assurance measure that will guide expectations for each treatment site. It will also allow creation of guidelines to determine acceptable shifts between the spine and a given target (fiducial) for a given site. Estimating the magnitude of this adjustment for pancreatic cancer stereotactic body radiation therapy (SBRT) was the purpose of this work. This was a retrospective study of nineteen (n = 19) pancreatic SBRT patients, average age 65y (range 47 - 88y), treated between 12/2012 - 6/2013. Radio-opaque fiducial markers were implanted by endoscopic guidance into or adjacent to the pancreatic tumor to provide a target for accurate localization. For each fraction, a free-breathing CBCT was registered to a reference breath-hold CT for alignment to spine. Then, two perpendicular breath-hold kV projection images were acquired and compared with corresponding reference digitally reconstructed radiographs (DRR) to further fine-tune the alignment with the fiducial marker. By comparing the breath-hold kV projection images from subsequent treatment fractions with those from fraction 1, 3D variability of the fiducial position was derived. A 2D-to-2D image registration code was written that registered each day’s 2D projection image to the projection image from day 1. We applied normalized cross-correlation for this registration where the template was chosen at the spine. After fusing the two images at the spine region, the inter-fraction variability of the anatomical breath-hold state was then derived by analyzing the relative position of the fiducials from day n (n>1) compared to day 1 images. We observed an average inter-fraction reproducibility of 1.73 ± 0.84 mm, 1.98 ± 1.42 mm, and 3.19 ± 2.49 mm in the LR, AP and SI directions, respectively. The average excursion from free breathing spine alignment to breath-hold fiducial alignment was 1.51 ± 1.38 mm, 2.05 ± 1.91 mm, and 3.01 ± 2.04 mm in the LR, AP and SI directions, respectively. The range of excursions which can be interpreted as the average expected adjustment required after spine alignment was observed as 0.18 - 5.07 mm, 0.07 - 5.89mm, and 0.61 - 7.82 mm in the LR, AP and SI directions respectively. Our study shows that inter-fraction shifts from free-breathing spine to ABC breath-hold target can be as high as 8 mm. Values that deviate significantly from this limit should be investigated to rule out potential problems such as bowel gas or fiducial shift prior to proceeding with treatment.