Real-time motion tracking of a gold implant in water by Cherenkov lights

The intra-fraction motion of tumors during radiation therapy can be monitored using imaging systems based on X-ray imaging (e.g., kV/MV radiography or CBCT) or magnetic resonance imaging (MRI). However, these techniques have limitations, including additional ionizing radiation dose and the need for complex, expensive equipment. Hence, a non-invasive, lower-cost motion-tracking approach that does not add radiation would be beneficial for routine clinical applications.

To study the feasibility of tracking the dynamic motion of a small gold marker in water by using Cherenkov photons produced by MV photon beams, through Monte Carlo simulations and experiments.

The GAMOS Monte Carlo code, which includes an optical simulation module, was used to calculate the deposited dose and Cherenkov-light yield in a water phantom. Phase-space files for 6, 10, 15, and 10 MV flattening filter-free (FFF) beams served as photon sources. The simulation setup consisted of a 20 × 20 × 20 cm³ water phantom with an 8 × 12 × 0.25 mm³ gold planar implant at its center. Percent depth dose (PDD) and Cherenkov-light yield were evaluated throughout the phantom and near the gold implant. For experimental validation, a 20 × 20 × 20 cm³ glass water tank was placed on a programmable platform executing sinusoidal, sawtooth, and sharkfin motions (± 20 mm amplitude; 10- and 60-s cycles). A 5 × 5 cm² 10 MV FFF beam (2400 MU/min) irradiated the gold implant. A scientific CMOS camera (f/2.0) positioned at 80 cm from the implant captured Cherenkov images. Repeatability was quantified by the standard deviation (SD) across motion cycles. The accuracy was assessed as the absolute difference between the bright-room reference position and the Cherenkov-derived position. Images of the implant in sinusoidal motion were also acquired during delivery of a two-arc brain volumetric modulated arc therapy (VMAT) plan (Arc A and Arc B).

Monte Carlo simulations showed that Cherenkov light yield throughout the phantom closely correlated with the PDD for all beam energies. The Cherenkov light yield upstream of the gold implant increased by 6.12%, 7.16%, 9.11%, and 7.01% compared to the Cherenkov light yield without the gold implant for 6, 10 FFF, 10, and 15 MV beams, respectively. Experiments confirmed that the beam-facing side of the gold implant appeared brighter than the opposite side, indicating enhanced backscattered electron production. Cherenkov-based positional measurements of the periodically moving implant showed close agreement with bright-room references. For static beams, accuracy across motion patterns was < 0.70 mm with repeatability < 0.2 mm. For the VMAT plan, the accuracies were < 0.50 mm (Arc A) and < 0.70 mm (Arc B).

The Cherenkov-light yield was strongly correlated with the deposited dose and showed a notable enhancement around the gold implant. Cherenkov light imaging successfully tracked the dynamic motion of a gold implant in water under high-dose rate conditions, demonstrating its potential for real-time fiducial tracking in radiation therapy without additional radiation exposure.