Room: AAPM ePoster Library
Purpose: To quickly and efficiently acquire projected percent depth dose (pPDD) data and projected crossplane profiles (pCPP) to verify MR-linac output is well-modeled by the Monte Carlo-based TPS during commissioning.
Methods: Cherenkov images were acquired from a water tank irradiated with a variety of beams during commissioning of a 0.35T MR-linac. An intensified CMOS camera was mounted on the wall at a distance of 5.5 m to isocenter and time-gated to the linac pulses. A blacked-out 40 x 40 x 40 cm3 water tank was aligned with the water surface at 78 SSD. Various field sizes were used to measure pPDDs and pCPPs at multiple depths. GAMOS simulations were used to correct for nonlinearity between Cherenkov and dose due to beam hardening with increasing depth, and optical pPDDs and pCPPs were compared to the y-projected TPS dose volume.
Results: Agreement between Cherenkov and TPS pPDDs was greatly improved with the simulation-based beam hardening correction, giving an average increase in gamma pass rate at 2%/2mm from 7% to 94% for larger field sizes. However, readout noise lowers SNR, increasing discrepancies between TPS and optical pPDDs. Comparison of pCPPs shows strong agreement from the center to edge of each field, although issues arise in the penumbra due to noise and loss of CPE, which causes nonlinearity between Cherenkov emission and dose.
Conclusion: This method can greatly increase efficiency for MR-linac commissioning, as all data for one field is acquired on the order of 30 seconds, compared with around 5 minutes per data point with hand-actuated techniques. Cherenkov pPDDs and pCPPs show strong agreement with TPS data when SNR is adequate and beam hardening corrections are applied. Future work includes optimizing acquisition settings to improve SNR and improving simulations by using vendor-provided phase space files.
Funding Support, Disclosures, and Conflict of Interest: This project has been sponsored by National Institutes of Health research grants R01EB023909 and R44CA199681. The authors acknowledge the Irradiation Shared Resource at the Norris Cotton Cancer Center at Dartmouth with NCI Cancer Center Support Grant 5P30 CA023108-41. Brian W Pogue reports financial interest in DoseOptics LLC.