Assessment of spatial uncertainty in computed tomography-based Gamma Knife stereotactic radiosurgery process with automated positioning system.

BACKGROUND: In this study, we assessed the geometric accuracy of an automated positioning system in Gamma Knife (GK) surgery. Specifically, we looked at the total spatial uncertainty over the entire treatment range of GK stereotactic radiosurgery (SRS) procedures in both the GK model C and the Perfexion (PFX). METHODS: An originally-developed phantom and a radiochromic film were used for obtaining actual dose distributions. The phantom, with inserted films on different axial planes (z = 60, 75, 100, 125, 140 mm), sagittal planes (x = 60, 75, 100, 125, 140 mm), and coronal planes (y = 60, 75, 100, 125, 140 mm), was placed on a Leksell skull frame. Computed tomography (CT) was then performed with a stereotactic localizer box attached to the frame, and dose planning was made using the Leksell GammaPlan treatment planning system. The phantom finally received beam delivery using a single shot of a 4-mm collimator helmet. The discrepancy between the planned shot position and the irradiated center position was evaluated by a dedicated film analysis software. RESULTS: The total uncertainty of CT-based GK SRS was less than 1 mm for almost all measured points over the stereotactic space in both the model C and the PFX. In addition, the geometric accuracy of the automated positioning system was estimated to be less than 0.1 mm and equal to 0.5 mm in the central and peripheral areas, respectively. CONCLUSIONS: We confirmed that the total spatial uncertainties of both the GK model C and the PFX are acceptable for clinical use.

Dosimetric comparison of absolute and relative dose distributions between tissue maximum ratio and convolution algorithms for acoustic neurinoma plans in Gamma Knife radiosurgery.

BACKGROUND: The treatment planning for Gamma Knife (GK) stereotactic radiosurgery (SRS) that performs dose calculations based on tissue maximum ratio (TMR) algorithm has disadvantages in predicting dose in tissue heterogeneity. The latest version of the planning software is equipped with a convolution dose algorithm as an optional extra and the new algorithm is able to compensate for head inhomogeneity. However, the effect of this improved calculation method requires detailed validation in clinical cases. In this study, we compared absolute and relative dose distributions of treatment plans for acoustic neurinoma between TMR and the convolution calculation. METHODS: Twenty-nine clinically used plans created by TMR algorithm were recalculated by convolution method. Differences between TMR and convolution were evaluated in terms of absolute dose (beam-on time), dosimetric parameters including target coverage, selectivity, conformity index, gradient index, radical homogeneity index and the dose-volume relationship. RESULTS: The discrepancy in estimated absolute dose to the target ranged from 1 to 7 % between TMR and convolution. In addition, dosimetric parameters of the two methods achieved statistical significance. However, it was difficult to see the change of relative dose distribution by visual assessment on a monitor. CONCLUSIONS: Convolution, heterogeneity correction calculation, and the algorithm are necessary to reduce the dosimetric uncertainty of each case in GK SRS.

[Assessment of overall spatial accuracy in image guided stereotactic body radiotherapy using a spine registration method].

Stereotactic body radiotherapy (SBRT) for lung and liver tumors is always performed under image guidance, a technique used to confirm the accuracy of setup positioning by fusing planning digitally reconstructed radiographs with X-ray, fluoroscopic, or computed tomography (CT) images, using bony structures, tumor shadows, or metallic markers as landmarks. The Japanese SBRT guidelines state that bony spinal structures should be used as the main landmarks for patient setup. In this study, we used the Novalis system as a linear accelerator for SBRT of lung and liver tumors. The current study compared the differences between spine registration and target registration and calculated total spatial accuracy including setup uncertainty derived from our image registration results and the geometric uncertainty of the Novalis system. We were able to evaluate clearly whether overall spatial accuracy is achieved within a setup margin (SM) for planning target volume (PTV) in treatment planning. After being granted approval by the Hospital and University Ethics Committee, we retrospectively analyzed eleven patients with lung tumor and seven patients with liver tumor. The results showed the total spatial accuracy to be within a tolerable range for SM of treatment planning. We therefore regard our method to be suitable for image fusion involving 2-dimensional X-ray images during the treatment planning stage of SBRT for lung and liver tumors.