Leksell GammaPlan v10 allows for tissue heterogeneities using a convolution algorithm. 

     In this study we investigate the dosimetric implications of using this new algorithm.

     Data from 12 patients was studied.

     Twelve patient data sets comprising 19 treatment scenarios (tumors and functional disorders) were accessed for retrospective planning. Each scenario was planned using both the standard and new algorithms. Treatment time, target coverage and dose to critical structures were compared. A computer-simulated 8 cm radius spherical phantom with 7 mm thick hard bone at a depth of 10-17 mm was used to test the effect on single shots in a simple geometry. Finally, an effective depth calculation was performed on two of the patient data sets.

     The mean time to deliver the same prescription dose was 5.4%±2.1%(standard deviation) longer when heterogeneities were considered, range 2.1–8.8%. Despite the change in treatment time, the relative dose distributions appeared relatively unaffected by this change in algorithm. Parameters such as D100 and maximum and minimum dose to target changed by not more than ±1% on average. Nearby brainstem dose changed on average by -0.4±2.0% (range: -3.3%-4.0%). Other critical structures considered received too low a dose to make changes relevant. The single shot and effective depth calculations in phantom showed an increase in treatment time of 3.2–6.9%, depending on position. These results support the patient dosimetry well.

     This was a retrospective study.

     With the new convolution-based dose algorithm, treatment times for the same prescription dose increase on average by 5%, but the dose distribution remains relatively unaffected. This is consistent with accounting for the hard bone in the skull, as verified with a simple effective depth approach. 

     Dosimetric film measurements in an anthropomorphic head phantom with bone heterogeneities is planned to verify the new algorithm.