Premium
Evaluation of the gamma dose distribution comparison method
Author(s) -
Low Daniel A.,
Dempsey James F.
Publication year - 2003
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 - 10.1118/1.1598711
Subject(s) - gamma distribution , dosimetry , noise (video) , mathematics , generalized gamma distribution , distribution (mathematics) , physics , nuclear medicine , statistics , mathematical analysis , computer science , medicine , artificial intelligence , image (mathematics)
The γ tool was developed to quantitatively compare dose distributions, either measured or calculated. Before computing γ, the dose and distance scales of the two distributions, referred to as evaluated and reference, are renormalized by dose and distance criteria, respectively. The renormalization allows the dose distribution comparison to be conducted simultaneously along dose and distance axes. The γ quantity, calculated independently for each reference point, is the minimum distance in the renormalized multidimensional space between the evaluated distribution and the reference point. The γ quantity degenerates to the dose‐difference and distance‐to‐agreement tests in shallow and very steep dose gradient regions, respectively. Since being introduced, the γ quantity has been used by investigators to evaluate dose calculation algorithms, and compare dosimetry measurements. This manuscript examines the γ distribution behavior in two dimensions and evaluates the γ distribution in the presence of data noise. Noise in the evaluated distribution causes the γ distribution to be underestimated relative to the no‐noise condition. Noise in the reference distribution adds noise in the γ distribution in proportion to the normalized dose noise. In typical clinical use, the fraction of points that exceed 3% and 3 mm can be extensive, so we typically use 5% and 2–3 mm in clinical evaluations. For clinical cases, the calculation time is typically 5 minutes for a 1 × 1 mm 2interpolated resolution on an 800 MHz Pentium 4 for a 14.1 × 15.2 cm 2radiographic film.