SU-E-T-647: Plan Quality in Computerized Non-Coplanar IMRT Beam Angle Optimization is Highly Dependent on the Extent of the Beam direction Search Space.

Purpose: To investigate the relationship between plan quality and the extent of the beam direction search space in computerized beam angle selection for generating optimal (non-coplanar) IMRT plans for prostate SBRT with dose distributions simulating HDR brachytherapy.

Methods: iCycle (1) was used to investigate the relationship between plan quality and the extent of the set of beam directions available for plan generation. For a group of 10 prostate patients, optimal plans were generated for 5 direction search spaces. For coplanar treatments (CP set), 72 orientations were available for selection (separation 5°). The fully non-coplanar set (F-NCP) included the CP directions plus 430 directions spread over the sphere. The CK set contained the directions available at the robotic Cyberknife unit. CK+ and CK++ were extensions of CK to investigate some of its characteristics. Generated plans were in accordance with our clinical SBRT protocol for Cyberknife treatment, delivering 4 fractions of 9.5 Gy. Adequate PTV coverage had the highest priority. Reduction of rectum dose was the highest OAR priority.

Results: The mean PTV coverage (V95) of all SBRT plans was 99% ï,± 0.9% (1 SD). F-NCP plans had most favorable OAR dose parameters, while for coplanar plans OAR doses were highest. Compared to coplanar treatment, rectum Dmean/V60 were 25% / 37% and 19% / 21% lower in F-NCP and CK plans. Higher rectum dose for the Cyberknife set compared to F-NCP was not caused by a lack of posterior beams for Cyberknife. For all search spaces, reduction in OAR dose only leveled off with > 20 beams in the plans (for CP, rectum V60 in 25 beam plans was reduced by 64% compared to 11 beams). In the non-coplanar set-ups, there was a preference for beams with a (large) lateral component.

Conclusions: Plan quality clearly improved with the extent of the beam direction search space (coplanar worst), and the number of beam directions in the plan (25 clearly better than 11).(1) Breedveld S, Storchi P, Voet P, Heijmen B, Med Phys 2012; DOI: 10.1118/1.3676689.