Unencrypted login | home

Program Information

Validation of the Leksell GammaPlan Convolution-Based Dose Calculation Algorithm

B Hundertmark

B Hundertmark, Y Watanabe, University of Minnesota, Minneapolis, MN

SU-E-T-552 Sunday 3:00PM - 6:00PM Room: Exhibit Hall

Purpose: To validate the new convolution-based dose calculation algorithm within the Leksell GammaPlan 10.1.1 treatment planning system (TPS) and to compare the new algorithm with the TMR 10 classical algorithm.

Methods: Using both the convolution and the TMR 10 algorithms, two test cases, a single 18 mm shot plan (Rx = 10 Gy to the 50% isodose) and a plan with 4 overlapping 18 mm shots (Rx = 15 Gy to the 50%) were separately computed on the spherical phantom used for GammaKnife output calibration (diameter of 16 cm, density = 1.06 g/cm3). The dose at the center of the phantom was measured using a micro chamber and gafchromic film. The ISP FilmQA Pro program was used to compare the films with TPS-predicted planar doses.

Results: For the single-shot plan, the predicted mean dose to the ion chamber was 19.6 Gy and 19.9 Gy for convolution and TMR 10, respectively. The measured dose was 20.2 Gy, which was 3% higher than the convolution prediction and 1.5% higher than TMR 10 predicted. For the four-shot case, the predicted dose was 19.3 Gy for convolution and 19.6 Gy for TMR 10. The measured dose was 20.4 Gy (5.7% higher than convolution and 1.4% higher than TMR 10). For the single shot, gamma analysis (2% and 1 mm) produced passing rates of 99.8 % for convolution and 98.3% for TMR 10. For the four-shot plan, the passing rates were 99.8% and 96.8% for convolution and TMR10, respectively.

Conclusion: The convolution algorithm produced better gamma passing rates than TMR 10; however, convolution under-predicted absolute dose and its error was exacerbated with addition of shots. This can in part be explained by fact that the output calibration of the machine is done assuming that the sphere is composed of water, but further investigation is needed.

Contact Email: