A new Gamma Knife radiosurgery paradigm: tomosurgery

This study proposes and simulates an inverse treatment planning and a continuous dose delivery approach for the Leksell Gamma Knife (LGK, Elekta, Stockholm, Sweden) which we refer to as "Tomosurgery." Tomosurgery uses an isocenter that moves within the irradiation field to continuously deliver the prescribed radiation dose in a raster-scanning format, slice by slice, within an intracranial lesion. Our Tomosurgery automated (inverse) treatment planning algorithm utilizes a two-stage optimization strategy. The first stage reduces the current three-dimensional (3D) treatment planning problem to a series of more easily solved 2D treatment planning subproblems. In the second stage, those 2D treatment plans are assembled to obtain a final 3D treatment plan for the entire lesion. We created Tomosurgery treatment plans for 11 patients who had already received manually-generated LGK treatment plans to treat brain tumors. For the seven cases without critical structures (CS), the Tomosurgery treatment plans showed borderline to significant improvement in within-tumor dose standard deviation (STD) (p <0.058, or p <0.011 excluding case 2) and conformality (p < 0.042), respectively. In three of the four cases that presented CS, the Tomosurgery treatment plans showed no statistically significant improvements in dose conformality (p <0.184), and borderline significance in improving within-tumor dose homogeneity (p <0.054); CS damage measured by V20 or V30 (i.e., irradiated CS volume that receives > or =20% or > or =30% of the maximum dose) showed no significant improvement in the Tomosurgery treatment plans (p<0.345 and p <0.423, respectively). However, the overall CS dose volume histograms were improved in the Tomosurgery treatment plans. In addition, the LGK Tomosurgery inverse treatment planning required less time than standard of care, forward (manual) LGK treatment planning (i.e., 5-35 min vs 1-3 h) for all 11 cases. We expect that LGK Tomosurgery will speed treatment planning and improve treatment quality, especially for large and/or geometrically complex lesions. However, using only 4 mm collimators could greatly increase treatment plan delivery time for a large brain lesion. This issue is subject to further investigation.