(Sunday, 7/14/2019)
Room: ePoster Forums
Purpose: A major challenge in lung stereotactic radiotherapy is to track the motion of lung tumors while the treatment beam is being delivered. One solution is to calculate the 3D displacement vector of respiratory motion using MV and kV x-ray images that are acquired simultaneously during treatment at orthogonal gantry angles. To set up the baseline for the displacement vector, a high-resolution cone beam CT (CBCT) image of the patient must be acquired prior to the treatment.
Methods: High resolution CBCT imaging is achieved through optimized reconstruction algorithms and gated CBCT acquisition. ITK (Insight Segmentation and Registration Toolkit) and RTK (Reconstruction Toolkit) toolkits are both open-source, cross-platform systems that offers an extensive collection of libraries for image processing and fast cone beam CT reconstruction. Currently the ITK/RTK toolkits are set up on a workstation with GPU computing capability. The reconstruction kernel is optimized for high spatial resolution by modifying the filter window shape and cut-off frequency. High reconstruction noise is allowed, which can be reduced to clinically acceptable levels through forward projecting, since the succeeding tracking algorithm compares MV and KV images with forward projections of CBCT at arbitrary gantry angles. Gated CBCT acquisition is used with a respiratory chest phantom to eliminate motion artifacts caused by breathing. Data acquisition is performed on a Varian TrueBeam linear accelerator (v2.7) and gated at end-expiration phase with a range of 5%.
Results: The ITK/RTK-based reconstruction platform can reconstruct 3D volume images of a chest phantom with an isotropic voxel size of 0.5mm and a spatial resolution up to 1.0lp/mm. A contrast-to-noise ratio up to 3.0 can be achieved around tumor.
Conclusion: The ITK/RTK platform is able to reconstruct CBCT images with clinically acceptable resolution. Lung tissues around tumor are visible in forward projections, which serve as good landmarks for succeeding tracking algorithms.
Funding Support, Disclosures, and Conflict of Interest: Varian MRA
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email
