Comparison of five dose calculation algorithms in a heterogeneous media using design of experiment. (May 2019)

• Record Type: Journal Article
• Authors: Vangvichith, M.; Autret, D.; Tiplica, T.; Barreau, M.; Dufreneix, S.
• Is Part Of: Physica medica. Volume 61(2019)
• Page Start: 103
• Page End: 111
• Publication Date: 2019-05
• Subjects: Design of experiment -- Heterogeneous media -- Algorithm -- Dose calculation; Medical physics -- Periodicals; Biophysics -- Periodicals; Biophysics -- Periodicals; Imagerie médicale -- Périodiques; Radiothérapie -- Périodiques; Rayons X -- Sécurité -- Mesures -- Périodiques; Physique -- Périodiques; Médecine -- Périodiques; 610.153
• DOI: 10.1016/j.ejmp.2019.04.014
• Languages: English
• ISSNs: 1120-1797

Abstract

Highlights: Design of experiment offers a robust methodology to compare algorithms. Some type "b" algorithms are as good as type "c" algorithms. Deviations are mainly located in the heterogeneity inside the lung. Abstract: Purpose: Design of experiments (DoE) provides a methodology to reveal the influence of input values on the measured output with a limited number of trials. The purpose of this study was to describe how DoE can be used to evaluate the performances of several dose calculation systems in heterogeneous media, including algorithms like Pencil Beam (PB), Anisotropic Analytical Algorithm (AAA), Acuros XB (AXB), Monte Carlo (MC) and Collapsed Cone Volume (CCV). Method: This study was carried out using a CIRS Model 002LFC IMRT Thorax Phantom customized with a water-equivalent heterogeneity inside the lung. The calculated dose distributions were compared to Gafchromic® EBT3 film measurements. The beam configurations were selected using DoE to study the influence of five parameters simultaneously (energy, collimator angulation, gantry angulation, X and Y jaws) and to optimize the number of experiments. An analysis of variance was performed over the entire irradiation field and over various regions of interest (tumour, shadow of tumour and lungs). Results: DoE enabled to quantify and determine the statistically significant factors, leading to an evaluation of the dose calculation systems in the lung case. The resulting scoring could be as follow (from best to worst): AXB_Dm, CCV, AXB_Dw, XVMC_Dm, XVMC_Dw, AAA and last PB. Differences between the algorithms were specially observed in the tumour and the shadow regions. Conclusion: DoE is a robust statistical method to compare several dose calculation systems. The various analyses lead to the conclusion that AXB handled more accurately most of the situations investigated in heterogeneous media.