Click here to


Are you sure ?

Yes, do it No, cancel

Evaluating Patient Setup Accuracy Using a Subset of CBCT Projections in Fiducial Tracking of Abdominal SBRT

O Oderinde1*, H Mostafavi2, D Simpson1, J Murphy1, L Cervino3, G Kim1, (1) University of California San Diego, La Jolla, CA, (2) Varian Medical System, Palo Alto, CA, (3) Memorial Sloan Kettering Cancer Center, New York, NY


(Sunday, 7/12/2020)   [Eastern Time (GMT-4)]

Room: AAPM ePoster Library

Purpose: To show that tracking only a subset of the projection images acquired for pre-treatment CBCT can be more effectively used for positioning the patient. The less noisy subset are the projections acquired near AP and PA kV imaging directions.

Methods: Short anteroposterior arc beam (SAPAB) projection angles of 0-450, 135-2250, and 315-3600 were extracted from the pre-treatment CBCT projections of 8 pancreatic and 5 liver SBRT patients (41 treatment fractions) with implanted fiducials. The fiducials were tracked in both the SAPAB and full-CBCT (0-3600) projections using an offline research software (RapidTrack, Varian). The tracking results were compared based on (1) fiducial and external breathing signal correlation, (2) internal tracking margin between 5th and 95th percentiles in each cardinal direction and, (3) predicted couch shift for clinical decision, which also compared with the actual clinical decision.

Results: The average correlation coefficient (coeff) (±SD) for full and SAPAB CBCT projections were 0.93±0.07 and 0.95±0.07, respectively, with root mean square errors (RMSEs) between 0.01 and 1.08 mm. These RMSEs could cause the mean fiducial motion error between full and SAPAB CBCT projections up to 0.07±0.67, 0.19±1.32 and 0.22±1.17 mm in LAT, SI and AP directions, respectively. Comparing the tracking results from full projections with SAPAB projections, the mean predicted couch shifts deviate by 0.08±0.49, 0.32±1.25, and 0.08± 0.27 mm in LAT, SI, and AP directions. In this study, the predicted couch shifts agreed with the clinical couch shift with a mean deviation of 0.88±0.72 mm.

Conclusion: This study has shown that the internal-external correlation coefficient is improved in the absence of lateral arc CBCT projections, which also influence the motion margins and predicted couch shifts. However, the full CBCT projection is needed for producing the 3D image that provides positional information about the organs at risk as well as the tumor.

Funding Support, Disclosures, and Conflict of Interest: This study was funded by Varian Medical Systems, Palo Alto, California, USA.


Not Applicable / None Entered.


Not Applicable / None Entered.

Contact Email