Geometrical accuracy of the Novalis stereotactic radiosurgery system for trigeminal neuralgia.

OBJECT: Stringent geometrical accuracy and precision are required in the stereotactic radiosurgical treatment of patients. Accurate targeting is especially important when treating a patient in a single fraction of a very high radiation dose (90 Gy) to a small target such as that used in the treatment of trigeminal neuralgia (3 to 4-mm diameter). The purpose of this study was to determine the inaccuracies in each step of the procedure including imaging, fusion, treatment planning, and finally the treatment. The authors implemented a detailed quality-assurance program.
METHODS: Overall geometrical accuracy of the Novalis stereotactic system was evaluated using a Radionics Geometric Phantom Chamber. The phantom has several magnetic resonance (MR) and computerized tomography (CT) imaging-friendly objects of various shapes and sizes. Axial 1-mm-thick MR and CT images of the phantom were acquired using a T1-weighted three-dimensional spoiled gradient recalled pulse sequence and the CT scanning protocols used clinically in patients. The absolute errors due to MR image distortion, CT scan resolution, and the image fusion inaccuracies were measured knowing the exact physical dimensions of the objects in the phantom. The isocentric accuracy of the Novalis gantry and the patient support system was measured using the Winston-Lutz test. Because inaccuracies are cumulative, to calculate the system's overall spatial accuracy, the root mean square (RMS) of all the errors was calculated. To validate the accuracy of the technique, a 1.5-mm-diameter spherical marker taped on top of a radiochromic film was fixed parallel to the x-z plane of the stereotactic coordinate system inside the phantom. The marker was defined as a target on the CT images, and seven noncoplanar circular arcs were used to treat the target on the film. The calculated system RMS value was then correlated with the position of the target and the highest density on the radiochromic film. The mean spatial errors due to image fusion and MR imaging were 0.41+/-0.3 and 0.22+/-0.1 mm, respectively. Gantry and couch isocentricities were 0.3+/-0.1 and 0.6+/-0.15 mm, respectively. The system overall RMS values were 0.9 and 0.6 mm with and without the couch errors included, respectively (isocenter variations due to couch rotation are microadjusted between couch positions). The positional verification of the marker was within 0.7+/-0.1 mm of the highest optical density on the radiochromic film, correlating well with the system's overall RMS value. The overall mean system deviation was 0.32+/-0.42 mm.
CONCLUSIONS: The highest spatial errors were caused by image fusion and gantry rotation. A comprehensive quality-assurance program was developed for the authors' stereotactic radiosurgery program that includes medical imaging, linear accelerator mechanical isocentricity, and treatment delivery. For a successful treatment of trigeminal neuralgia with a 4-mm cone, the overall RMS value of equal to or less than 1 mm must be guaranteed.