Objectives: The purpose of this study is to examine the possibility of using the projections acquired for pre-treatment CBCT for patient set-up and motion tracking instead of 2D fluoroscopy for decision making in respiratory gating stereotactic body radiotherapy treatment.

Methods: Template matching and triangulation (RapidTrack offline research software, Varian) was used to track the fiducials in the fluoroscopic and CBCT images. We validated the tracking accuracy of the offline research software by comparing it's tracking result with known fiducial motion in a QuasarTM multipurpose body phantom and respiratory motion platform.  This retrospective study collected  pre-treatment CBCT and orthogonal fixed gantry fluoroscopic projections of ten abdominal  cancer cases (4 pancreatic and six liver) with implanted fiducial markers. The CBCT projections, the pre-treatment fluoroscopic images, and the external breathing signal, from a total of thirty-seven fractions, along with the treatment planning CT scan, were used for the study. Afterward, tracking results from the CBCT images were compared with fluoroscopic evaluated outcomes in terms of their external-internal motion modeling (correlation coefficient) and tumor motion. Pearson correlation coefficients were calculated to evaluate the internal-external correlation, and Mann-Whitney significant values were used to measure the significant difference between the CBCT and fluoroscopic data.

Results: From the phantom study, the template matching algorithm agrees with physical and moving images (from fluoroscopic projections) measurements in SI displacement within ±0.01 mm. Correlation coefficients in Superior-inferior (SI) direction for CBCT imaging were 0.95±0.20 and 0.91±0.21 for pancreas and liver cases, respectively. While in AP and LAT fluoroscopic projections,  correlation coefficients in SI were 0.98±0.13 and 0.97±0.18 for pancreas, and 0.97±0.12 and 0.97±0.11 for liver lesions, respectively. In CBCT, AP and LAT fluoroscopic projections, the average fiducial motion in 3D ranges were 12.40±0.83 mm, 9.89±1.94 mm, and 11.28±3.15 mm, respectively for pancreatic cases, while in liver cases, tracked fiducial motion ranges were 13.06±2.31 mm, 10.63±1.19 mm, and 8.37±4.49 mm. Using the CBCT projections for respiratory gating decision, tracking of the fiducial motions are statistically significant (P<0.05) in 81.5% and 70.2% treatment fractions of pancreas and liver lesions.

Conclusions: CBCT has the potential of being used for patient positioning, to the advantage of the patients in terms of reducing treatment time and imaging dose, thus replacing fluoroscopic projections.