Room: Exhibit Hall
Purpose: There have been large systematic differences between pencil-beam convolution, adaptive convolve, an analytical anisotropic algorithm (AAA) and Clarkson method for lung and breast, especially with the AAA showing a particularly large systematic difference of 8.7% for lung stereotactic body radiotherapy. In this study, we evaluated the impact of the CT-HU value in the voxels around the reference point.
Methods: 1502 clinically approved radiotherapy treatment plans were collected in our institution. Dose differences between three treatment planning systems and an independent calculation verification program using a Clarkson algorithm were computed in the plans. In addition to it, mean CT-HU value was computed using a cube region of interests (4 x 4 x 4 cmÂ³) where the reference point was located at the center. Both the dose differences and the mean CT-HU values were analyzed using the cluster analysis using k-means method.
Results: The cluster analysis using k-means method resulted in -300 CT-HU as the threshold value for the dose difference. Dose differences for the two groups which the CT-HU values were more than and less than -300 CT-HU were 0.4 Â± 1.6% and 4.1 Â± 2.5% respectively. The comparison for the two groups shows larger systematic and random differences with statistically significant difference (p < 0.01).
Conclusion: -300 CT-HU may be a prognostic factor for larger disagreement between the TPSs and independent calculation verification. Use of the heterogeneity correction for Clarkson method would be helpful when the CT-HU values were less than -300 CT-HU.