Encrypted login | home

Program Information

Light-Field Based Dynalog Verification

no image available
P Atwal

P Atwal*, R Ramaseshan , BC Cancer Agency, Abbotsford, BC

Presentations

SU-F-T-463 (Sunday, July 31, 2016) 3:00 PM - 6:00 PM Room: Exhibit Hall


Purpose: To independently verify leaf positions in so-called dynalog files for a Varian iX linac with a Millennium 120 MLC. This verification provides a measure of confidence that the files can be used directly as part of a more extensive intensity modulated radiation therapy / volumetric modulated arc therapy QA program.

Methods: Initial testing used white paper placed at the collimator plane and a standard hand-held digital camera to image the light and shadow of a static MLC field through the paper. Known markings on the paper allow for image calibration. Noise reduction was attempted with removal of ‘inherent noise’ from an open-field light image through the paper, but the method was found to be inconsequential. This is likely because the environment could not be controlled to the precision required for the sort of reproducible characterization of the quantum noise needed in order to meaningfully characterize and account for it. A multi-scale iterative edge detection algorithm was used for localizing the leaf ends. These were compared with the planned locations from the treatment console.

Results: With a very basic setup, the image of the central bank A leaves 15-45, which are arguably the most important for beam modulation, differed from the planned location by [0.38±0.28] mm. Similarly, for bank B leaves 15-45 had a difference of [0.42±0.28] mm

Conclusion: It should be possible to determine leaf position accurately with not much more than a modern hand-held camera and some software. This means we can have a periodic and independent verification of the dynalog file information. This is indicated by the precision already achieved using a basic setup and analysis methodology. Currently, work is being done to reduce imaging and setup errors, which will bring the leaf position error down further, and allow meaningful analysis over the full range of leaves.


Contact Email: