Program Information
A Dosimetric Validation of a Model-Based Treatment Planning Algorithm in Lung SBRT
A Meacham*, W Luo , J Molloy , University of Kentucky, Lexington, KY
Presentations
PO-BPC-Exhibit Hall-24 (Saturday, March 7, 2015) Room: Exhibit Hall
Purpose:
Dose at lung-tumor interfaces is not well-characterized by model-based treatment planning algorithms due to electronic disequilibrium. The interface region comprises a relatively large fraction of the total target volume in small lung nodules. This work evaluates the resulting dosimetric uncertainty in stereotactic body radiotherapy of small lung nodules.
Methods:
A Monte Carlo algorithm (MCSIM) based on EGS4 was used to validate dose calculations using the XiO superposition algorithm (CMS). Six lung SBRT plans consisting of 10 or 11 static, non-coplanar beams and treated on a Varian 21EX linac were evaluated. GTVs ranged from 0.96 to 10.98 cm³ in volume and varied in location within the lung. Planning and treatment parameters were held constant between the XiO and MCSIM plans. Electronic disequilibrium occurring at the lung-tumor interface was investigated by contouring a 2-mm ‘peel’ on the outer surface of the GTV. Clinical beam arrangements were studied, as were hypothetical arrangements including small numbers of beams and varied beam energies including 6 and 18 MV. Doses were prescribed to the 80% isodose line. Comparisons of DVHs, mean, minimum and maximum doses for the GTV, PTV and corresponding ‘peel’ structures were compared.
Results:
Mean doses in the GTV agreed to within 2.4% and 3.3% for the original 6 and 18 MV plans, respectively. For one patient the mean GTV dose for a one-beam plan was within 3% while the 10-beam arrangement showed an agreement of 1%. Mean doses in the GTV and 2-mm GTV peel were within 1.5% and 2.5% for both XiO and MCSIM for original 6 and 18 MV plans, respectively.
Conclusion:
XiO superposition performs accurate dose calculation in lung SBRT treatment planning in which a large number of static beams are used. Lower energy (6 MV) is preferable in terms of better agreement to Monte Carlo simulation.
Contact Email: