Premium
WE‐C‐BRB‐08: Sensitivity of Dosimetric Margin Distribution and Coverage Estimates to Sampling Parameters
Author(s) -
Xu H,
Gordon J,
Siebers J
Publication year - 2009
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.3182466
Subject(s) - standard deviation , dosimetry , nuclear medicine , mathematics , margin (machine learning) , sensitivity (control systems) , azimuth , range (aeronautics) , beam (structure) , statistics , physics , geometry , medicine , optics , materials science , computer science , machine learning , electronic engineering , engineering , composite material
Purpose: The dosimetric margin distribution (DMD) is a planning tool for evaluating target and normal tissue dose coverage probabilities in the presence of geometric uncertainties. This work evaluates sensitivity of coverage estimates derived from the DMD to the angular increment w at which the DMD is sampled, and the distance search increment d. Method and Materials: The study utilized three plans for localized prostate cancer, with prescribed dose of 79.2Gy and 1cm clinical target volume to planning target volume (CTV‐to‐PTV) margins, planned according to the high dose arm of RTOG0126. Dose was first recalculated with beam fluences convolved with a normal probability density function to simulate the effect of random uncertainties. Dosimetric margins were then calculated between the CTV and the 79.2Gy isodose surface at elevations f from −90° to +90°, and azimuthal angles q from −180° to +180°. In each direction (f,q) the CTV was moved relative to the dose distribution in steps of d until its minimum dose fell below 79.2Gy. Estimated coverage Qest was calculated from the DMD assuming systematic uncertainties (rigid body translations) were normally distributed with standard deviation S along each axis. Minimum and maximum deviations DQest were obtained over a range of S values for w = 5°, 10°, 20° relative to w = 2°, and for d = 0.2, 0.5, 1mm relative to d = 0.1mm. Results: Coverage uncertainty due to finite angular sampling interval w is approximately 0.5% at w = 10° and 1.5% at w = 20°. Uncertainty due to finite step size d is negligible for d £ 0.2 mm, and is approximately 0.6% for d = 1mm. Conclusion: Coverage estimates accurate to ±0.5% will be obtained if the DMD is sampled with w = 10° and d = 0.2 mm. Coverage estimates accurate to this level are desirable for target and normal tissue coverage evaluation in current treatment plans, and to support future treatment planning based on probabilistic coverage criteria. (Supported by NIH P01CA116602).