Toward planning target volume margin reduction for the prostate using intrafraction motion correction with online kV imaging and automatic detection of implanted gold seeds

Abstract

Purpose: The imaging application Auto Beam Hold (ABH) allows for the online analysis of 2-dimensional kV images acquired during treatment. ABH can automatically detect fiducial markers and initiate a beam interrupt. In this study, we investigate the practical use and results of this intrafraction monitoring tool for patients with prostate cancer who have implanted gold seeds treated with a RapidArc technique.

Methods and materials: A total of 105 patients were included. For setup, the seeds were lined up using 2 orthogonal 2-dimensional kV images. After the setup procedure, ABH was applied at an interval of 3 seconds. The software requires a maximum-allowed deviation to be defined for each seed, which is referred to as a deviation limit (DL). Online, the ABH application evaluates the position of the seeds and indicates for each seed whether or not it exceeds the DL. Patients were divided in 3 groups. For the first group ABH was used with the DL at 6 mm, which corresponds to the planning target volume (PTV) margin. For the second group, the DL was set at 5 mm with an unchanged PTV margin of 6 mm. For the third group, the PTV margin was reduced to 5 mm with a DL of 5 mm. Offline, we performed an analysis of the number of beam stops and resulting re-setups.

Results: ABH initiated a beam interrupt 223 times (13%) during a total of 1736 sessions. By decreasing the DL from 6 mm to 5 mm, the amount of workload for re-setups increased from 6% (group 1) to 14% (groups 2 and 3). Re-setup, 3-dimensional shifts larger than the PTV margin were found in 44%, 35%, and 45% for groups 1,2, and 3, respectively.

Conclusions: Intrafraction imaging of prostate position during treatment using automatic detection of implanted gold seeds was successfully implemented. PTV margins were safely reduced from 6mm to 5mm without a substantial increase in workload.