SU-E-T-406: Online Image-Guidance for Prostate SBRT: Dosimetric Benefits and Margin Analysis

To evaluate the dosimetric benefits of online image guidance during prostate stereotactic body radiotherapy (SBRT) and the potential on margin reduction.

28 prostate SBRT patients were retrospectively studied, each treated with 37Gy in 5 fractions. RTOG recently opened a similar protocol (0938). During treatments, per-beam couch corrections were made based on the actual target motion provided by dynamic tracking with either Calypso or per-beam OBI imaging. Dosimetric benefits of online correction were evaluated by comparing delivered dose distributions with and without such correction. The dose distribution without correction was generated in the same treatment planning system by accumulating doses without online correction from the each beam and each fraction. Quantitative analyses include the dosimetric difference between delivered doses with and without correction; the correction magnitude and frequency; and the potential on margin reduction based on the margin recipe by Van Herk et al.

(1) Delivery without online correction results in small reduction on target mean dose (0.03±0.05Gy), maximal dose (0.01±0.06Gy), and conformity index (<0.06). (2) Delivery without online correction has small impact on OAR dose: 26 out of 28 patients have <1%/1.5cc differences in V18.5Gy/V24Gy/V28Gy/V33Gy/V37Gy for both the bladder and the rectum. Maximal differences are 4cc of the bladder and 1.6cc of the rectum in mid-dose regions (V18.5Gy). (3) For femoral heads, <1cc/1Gy differences are observed in V20Gy/Dmean/D1cc.(4) Average number of couch corrections per fraction is 0.49. The magnitudes are: (-0.2±2)mm vertically, (-0.1±2.1)mm longitudinally, and (-0.2±1.4)mm laterally. (5) Margin determined by actual target motion in this patient population is 2.5mm isotropic.

For both target coverage and OAR sparing, overall small benefit is seen from per-beam couch correction under dynamic tracking. The target motion between beams is small and random, and indicates a population-based margin size of 2.5mm.