Comparing dose calculation algorithms in low dose regions for patients with brain metastases
PO-1400
Abstract
Comparing dose calculation algorithms in low dose regions for patients with brain metastases
Authors: ALAHMARI|, Mohammed(1)*[fehran82@hotmail.com];Wolfs|, Cecile(1);Temel|, Yasin(2);Eekers|, Danielle(1);Verhaegen|, Frank(1);
(1)GROW – School for Oncology and Developmental Biology- Maastricht University Medical Centre, Department of Radiation Oncology Maastro, Maastricht, The Netherlands;(2)School for Mental Health and Neuroscience- Faculty of Health- Medicine and Life Sciences- Maastricht University Medical Centre, Department of Neurosurgery, Maastricht, The Netherlands;
Show Affiliations
Hide Affiliations
Purpose or Objective
Dose calculation algorithms are used to precisely calculate the dose distribution to the target volumes and surrounding tissues. However, uncertainties still exist in dose distributions and usually occur beyond the high dose area of the planning target volume (PTV). This may increase the risk of radiation-induced side effects. The analytical anisotropic algorithm (AAA) and the Acuros XB algorithm are both in the Eclipse treatment planning system (TPS) for volumetric modulated arc therapy (VMAT). To the extent of our knowledge, studies comparing the two algorithms in the brain area are limited to three studies of which two were conducted on phantoms, and one on clinical data with only a small subgroup of neuro patients with no specific focus on low dose regions. Therefore, this study aimed to explore in silico the dose differences of these algorithms with a focus on low dose regions in brain metastases patients.
Material and Methods
After obtaining the ethics approval and by using a retrospective design, ten neuro metastases patients were loaded to the TPS for the purpose of comparison. For each patient, the 3D dose distributions were re-calculated by both algorithms; AAA (version 15.5.11) and Acuros XB (version 15.5.11), using the pre-set monitor unit values (MU) of the original treatment plans. Since we could not re-optimize the AAA before calculating, the AAA doses were normalized to the Acuros ones using the mean dose in the PTV. A comparison between the dose volume histograms (DVHs) of the two algorithms was then performed to extract the mean and maximum doses for the relevant organs-at-risk (OARs), namely: brain-PTV, brainstem, spinal cord, right and left cochlea, right and left optic nerve, right and left retina. The agreement between the two algorithms was assessed by gamma analysis (MATLAB), and the Bland Altman (B-A) test (SPSS). Gamma criteria were set to 1 mm distance to agreement and 2% dose difference.
Results
The gamma pass rates of the ten-neuro patients show 100% similarity between the two algorithms in all but three of the nine OARs. The brain-PTV, brainstem and the right cochlea yielded rates equal to 99.21%/ SD= 0.96, 99.87%/ SD= 0.36, and 99.48/ SD= 1.65, respectively. The scatter plot and sixteen out of the eighteen B-A plots (nine OARs * two DVH metrics (mean and max doses)) show a good agreement at lower dose levels, however, the profile of points exhibit an increasing trend with dose within the tolerance limits (figure 1 and 2).
Figure 1: A scatter plot of one algorithm against the other comparing the mean absolute doses to the line of perfect agreement.
Figure 2: A B-A plot of the right cochlea (the horizontal line of zero= no difference)
Conclusion
The agreement level between the algorithms is high and almost all of the differences in each OAR are located well within the default tolerance limits, suggesting that AAA and Acuros could be used interchangeably in neuro patients for low dose regions. Future work may include adding more patients for stronger conclusion.