Artikel
Evaluation of a new full convex optimisation algorithm for inverse planning in Gamma Knife radiosurgery
Bewertung eines neuen vollständig konvexen Optimierungsalgorithmus für die inverse Planung bei der Gamma-Knife-Radiochirurgie
Suche in Medline nach
Autoren
Veröffentlicht: | 25. Mai 2022 |
---|
Gliederung
Text
Objective: For the Gamma Knife radiosurgery (GKRS), a new inverse planning software (IP) based on a global convex optimization algorithm is adopted. The objective of the current study was to investigate the applications of IP for pre-planning in daily clinical use and its performance for different cerebral entities.
Methods: IP was tested in a prospective user study for a total of 117 patients and 230 target volumes. Conventional expert pre-plans were used as reference plans, which included forward planning by expert planners and subsequent local internal optimization. Based on the same dose constraints as the expert plans, the real-time calculation of the inverse calculated treatment plans was performed with its default optimization settings. To compare both planning techniques, hypothesis tests on radiosurgery plan quality metrics such as the Paddick conformity index (PCI) were performed whereby the three selectable IP planning strategies and the different entity groups were distinguished.
Results: The statistical tests revealed that IP-calculated treatment plans were almost equal in quality compared to pre-plans from expert planners. Especially IP plan quality metrics related to conformity and dose gradient attained statistically significant higher scores combined with high coverage except for the selective optimizing strategy. The other parameters examined such as normal brain V12 Gy or V10 Gy did not show notable differences for the two remaining strategies. Inversely generated plans displayed significantly shorter planning times versus (manual) expert plans as well as a greater number of isocenters, which often eventuates in longer treatment times. But regarding the total time, these differences are vanishingly slight.
Conclusion: The most important clinically relevant finding was that the use of IP is especially applicable for complex and irregular-shaped tumors but rather unsuitable for metastases. We observed clinical significance for IP in the conformity parameters and superiority for the strategy maximizing the selectivity. Overall, high-quality plans from IP enable newcomers to achieve almost equal qualitative treatment pre-plans. On the one hand, it allows optimizing the sparsity for surrounding tissue and the conformity for benign tumors in a short time frame. On the other hand, it forms a solid basis for further planning on the treatment day.