(Tuesday, 7/16/2019) 10:30 AM - 11:00 AM
Room: Exhibit Hall | Forum 9
Purpose: We previously developed a segmentation based semi-automatic sliding motion compensated 4D-CBCT reconstruction scheme. However it breaks the image reconstruction frequently since image segmentation was needed during reconstruction iterations. This study investigates a bilateral filtering based 4D-CBCT scheme for sliding motion modeling, and make the whole 4D-CBCT reconstruction fully automatic.
Methods: Initially, a modified simultaneous algebraic reconstruction technique is used to generate a high quality 0% phase using all phase projections. Then the 4D-DVF is estimated by matching the measured projection of the target phase with the simulated forward projection of the deformed 0% phase CBCT. The sliding motion is modeled by performing a bilateral filtering on the DVF during optimization. The filter kernel contains: 1) an image spatial domain Guassian kernel; 2) an image intensity domain Guassian kernel; and 3) a DVF domain Guassian kernel. By choosing suitable kernel variance parameters, sliding motion was automatically extracted. A non-linear conjugate gradient optimizer was used. We tested the algorithm on a non-uniform rotational B-spline based cardiac-torso (NCAT) phantom and two anonymized patient data. For quantification, the Root-Mean-Square-Error (RMSE)/the Maximum-Error (MaxE) and the Dice coefficient of the extracted lung contour from the final reconstructed images were used for comparison.
Results: NCAT phantom’s motion trajectory's RMSE and MaxE are 0.796 mm and 1.02 mm for bilateral filtering based reconstruction. Compared with that without bilateral filtering, the trajectory's RMSE and MaxE are 2.704 mm and 4.08 mm, respectively. For the patient pilot study, the 4D-Dice coefficient obtained with bilateral filtering are consistently higher than that without bilateral filtering. Meantime the chest rib position and heart edge definition has been better corrected with bilateral filtering.
Conclusion: Bilateral filtering based fully automatic 4D-CBCT reconstruction scheme achieves sub-mm range reconstruction and motion tracking error. It’s hence a useful 4D-CBCT image guidance tool for radiation therapy.
Funding Support, Disclosures, and Conflict of Interest: This research is supported by Varian and a research grant from Chongqing Municipal People's Society Bureau
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email
