Investigation on the TMR10 and convolution algorithms of the gamma plan treatment planning system utilizing the lightning software for the treatment of vestibular schwannoma in the gamma knife icon unit

Abstract

Introduction: Leksell Gamma Plan (LGP) treatment planning is a software system that offers two-dose calculation algorithms for stereotactic radiosurgery/stereotactic radiotherapy treatments. The latest version of LGP is based on lightning software and includes TMR10 and convolution algorithms. TMR10 considers all brain structures as water and calculates the dose based on attenuation in water, while convolution accounts for the heterogeneous tissue inside the brain based on electron densities of all structures in the planning CT images. This study compares the dosimetric outcomes of these two algorithms for vestibular schwannoma cases and evaluates the impact of lightning software in heterogeneity corrections. Materials and Methods: Our Institute has conducted a comprehensive analysis of data from 16 patients who were diagnosed with vestibular schwannoma and underwent treatment with LGK SRS plans utilizing LGP Version 11.3 lighting software. These plans were developed using both TMR10 and convolution algorithms, with a prescribed dose of 12 Gy to the planning target volume (PTV) and a 50% isodose curve utilized to effectively cover the PTV while minimizing potential harm to healthy tissue. Our analysis focused on assessing the dosimetry data for the treatment plans, specifically in relation to the brainstem and cochlea as organs at risk. In our study, we conducted a comparison of the treatment plan indices for TMR and convolution using paired t-test analysis. Results: The maximum dose of PTV was 22.1 (1.7) Gy in the TMR10 algorithm and 21.4 (2.0) Gy in the convolution algorithm. The mean dose was 16 Gy and 15.2 Gy for TMR10 and convolution, respectively, with a difference of –5.00%. The Paddick conformity index was 0.82 and 0.84 for TMR10 and convolution, respectively, with a difference of 2.43%. The maximum and mean doses of the brainstem were 10.1 Gy and 9.8 Gy, and 2.6 Gy and 2.5 Gy in TMR10 and convolution algorithms, respectively. Conclusion: The TMR10 algorithm assumes all structures are unit densities and is traditionally used in LGK units for dose calculation. However, the convolution algorithm accurately predicts the delivered dose by accounting for in-homogeneities in the head structure. The use of a convolution algorithm may be necessary when the target is surrounded by highly tissue in-homogeneities such as air tissue interface and bone structures. Additionally, it may be necessary for patients with anterior venous malformation who require embolization to arrest bleeding. The use of a convolution algorithm in clinical practice needs to be validated before implementation.