Room: AAPM ePoster Library
Clinical evidence for gamma knife radiosurgery (GKRS) is based on the TMR10-algorithm, which approximates the entire skull as water. Our purpose is to recalculate actual delivered plans with convolution-algorithm (CA), taking tissue heterogeneity into account, to link clinical outcome to more accurately calculated dose distributions. Previously this has been done by investigating treatment time differences, which is only interesting out of a management point of view but fails to show correlations with clinical outcome and calculated dose distributions. To explore the impact of tissue heterogeneity, the cohort consisted of Vestibular Schwannoma (Ethical permit EPN 2017/1760-31/1).
GKRS-treated patients were recalculated by exporting CA dose distributions from individual shots and summing these externally, weighted according to delivered shot-time. The resulting dose distributions calculated with TMR10 and CA were compared. For the CTV; D9, coverage, mean-, and max-dose were scored and compared. For Cochlea and Brainstem mean and max doses were evaluated. As a biproduct, the impact of determining the body contour with the 16-points measurement bubble and from MRI images was evaluated. Significance was evaluated using the Wilcoxon’s signed rank test.
All doses calculated with CA are significantly lower than TMR10 (p<0.05). The D95 (coverage) was decreased with 9.2±1.6% (8.7±4.3%) consisting of 2.7±1.0% (1.7±0.9%) from different Body contours and 6.7±1,1% (7.4±3.7%) from the different algorithms. For Cochlea and Brainstem, mean doses are decreased with 12.9±1.7% and 7.2±0.7%, respectively, where 3.2±2.1% and 2.3±0.4% are due to different Body contours and 10.0±2.0% and 5.0±0.7% due to algorithms, respectively.
Our study shows that doses calculated taking heterogenous media into account is significantly lower than those which current clinical evidence is based on. To achieve equivalent clinical effectiveness and sparing OARs, dose prescriptions accounting for heterogeneous media should be adjusted.