Exhibit Hall | Forum 5
Purpose: Differential motion of multiple tumors can result in dosimetric error for patients with locally advanced cancer. To address this issue, a multileaf collimator (MLC) tracking method that corrected dose error in real time for independently moving targets was developed. This study investigated real-time multi-target tracking optimized using a fast dose calculation and compared the dosimetric accuracy to previous treatment methods.
Methods: A multi-target MLC (MT-MLC) tracking algorithm that optimized leaf positions in real time based on accumulated dose error was developed for moving prostate and static lymph node targets. To enable real-time dose accumulation, the dose calculation was simplified to deposit a unit of dose in the treatment volumes along the line-of-sight of the aperture at each timestep. The MLC leaves were then adapted to minimize the difference between the planned and delivered doses with motion. The MT-MLC tracking algorithm was evaluated by simulating treatment for five locally advanced prostate cancer patients and three motion traces. Delivered doses were evaluated using a clinically accepted dose calculation in a treatment planning system (TPS) (Eclipse v16.1). The performance of dose-optimized MT-MLC tracking was compared to a 2D geometric-optimized MT-MLC tracking approach, and no tracking, using a 2%/2mm gamma-pass criterion.
Results: Dose-optimized MT-MLC tracking had the lowest error with mean gamma-failure rates of 11.5%±8.5% for the prostate and 2.2%±3.2% for the nodes, compared to 22.6%±11.6% and 3.6%±2.5% for geometric-optimized MT-MLC tracking (p=0.02), and 37.1%±27.8% and 23.6±3.2% without tracking (p<0.01). The difference in dose errors for the fast dose calculation and the TPS dose ranged between -6.6% and 5.2% for 99% of voxels.
Conclusion: Though the fast dose calculation was simplified and did not account for all physical interactions of radiation dose in tissue, dose errors between the planned and delivered doses were sufficiently accurate to guide MLC optimization, enabling improved multi-target tracking.
Funding Support, Disclosures, and Conflict of Interest: The authors acknowledge funding provided from a Cancer Council NSW Project Grant (AP1165097). P.J. Keall is an inventor on US patents 7,469,035 and 8,971,489 that are related to MLC tracking. Patent 7,469,035 is unlicensed; patent 8,971,489 is exclusively licensed to Asto CT.
Not Applicable / None Entered.