Evaluation of Bony Anatomy Versus Endobiliary Stents as Surrogates for Volumetric Image Guidance in Pancreatic Cancer.

PURPOSE: Online treatment setup verification through cone-beam computed tomography (CBCT) in pancreatic cancer patients is limited by low soft tissue contrast. This study aims to quantify the relative positional displacements between bony anatomy and endobiliary stents as surrogates for pancreatic cancers. METHODS: Under ethics approval, 258 localization CBCT images from 15 pancreatic patients with endobiliary stents were evaluated. CBCTs were registered through two methods to assess translations and rotations: target adjacent bony anatomy through automatic registration and automatic stent registration through a shaped region of interest. Displacement vector differences between surrogate registrations were calculated and analysed. RESULTS: Mean (±standard deviation) translational displacements in the right/left, superior/inferior, anterior/posterior directions were 0.9 ± 3.1 mm, 1.8 ± 4.2 mm, and 0.4 ± 2.5 mm for bone registrations, respectively, and 0.9 ± 5.6 mm, -1.5 ± 5.7 mm, and -0.5 ± 4.3 mm for stent registrations, respectively. Mean (±standard deviation) rotational displacements for pitch, roll, and yaw were 0.16 ± 0.97°, -0.32 ± 0.96°, and -0.77 ± 1.8° for bone registrations, respectively, and -0.94 ± 4.6°, -0.4 ± 7.4°, and -0.13 ± 6.64° for stent registrations, respectively. Mean displacement vector between surrogates was 4 mm, with 43% of fractions measuring displacement vectors >5 mm. A maximum displacement vector of 22.6 mm between surrogates was observed. CONCLUSIONS: Varying positional differences were observed between bone and stent registration for pancreas CBCT-image-guided radiation therapy. Setup errors for stent matching were larger than bone registrations. Further research is required to determine if endobiliary stent position is equivalent to the pancreas' location to determine its suitability as a surrogate.