Click here to

Session: Therapy General ePoster Viewing [Return to Session]

A Comparison of Primary and Secondary Dose Calculation Algorithms for Lung SBRT Targets

J Pogue*, R Popple, J Stahl, R Sullivan, University of Alabama at Birmingham, Birmingham, AL


PO-GePV-T-421 (Sunday, 7/10/2022)   [Eastern Time (GMT-4)]

ePoster Forums

Purpose: While primary dose calculation algorithms have been studied in the presence of lung heterogeneity, agreement of these algorithms with robust secondary dose calculation algorithms is sparsely characterized. To evaluate this relationship, an analysis of SBRT lung target metrics was performed utilizing multiple clinically commissioned algorithms.

Methods: Twenty-five patients previously receiving lung SBRT treatment (50 – 55 Gy in 3 or 5 fractions) at our institution were utilized in this IRB-approved study. Tumor volumes were classified by location (central or peripheral based on RTOG protocols) and size (v < 30 cc and v ≥ 30 cc), then evaluated as a function of dose metric (mean, D95, and max) and algorithm. Primary dose calculation algorithms utilized in this work include the Analytical Anisotropic Algorithm (AAA 13.6.23) and Acuros XB (AXB 15.5.11), whereas a collapsed cone convolution algorithm (Mobius3D 4.0.1) was used for secondary calculation. When comparing primary and secondary algorithms, the number of monitor units was forced to match. For comparisons between AAA and AXB, plans were normalized so 95% of the target received prescription dose.

Results: The largest dose discrepancies were observed between AAA and Mobius, while AXB and Mobius demonstrated the best overall dose agreement. The Wilcoxon paired, non-parametric test was used to test algorithm induced dose metric differences. Significant differences include the D95 and Dmax differences between AAA and Mobius (p < 0.001 and p = 0.001, respectively), and the mean dose difference between AXB and Mobius (p < 0.001). Given equal normalization, AAA and AXB mean and max dose distributions are significantly different (p = 0.001 and p < 0.001, respectively).

Conclusion: A comparison of AXB and Mobius generally results in significantly smaller percent differences than AAA and Mobius. Agreement between primary and secondary algorithms should be well described to understand impacts of algorithm on target dosimetry.


Convolution/superposition, Collapsed Cone Superposition


TH- External Beam- Photons: extracranial stereotactic/SBRT

Contact Email