Purpose: A new method for stereotactic irradiation of spinal malignancies is presented, with evaluations of the theoretic and practical limitations of localization accuracy and the implementation of the method in swine.
Materials and methods: In a percutaneous procedure, a minimum of three small (1.7-mm-diameter) titanium markers are permanently affixed to a vertebra. Markers are localized on biplanar radiographs while isocenter positions are determined on CT. An external fiducial frame defines a three-dimensional coordinate system through the patient. Radiographs coupled with a rigid body rotation algorithm account for daily differences in patient position. Phantom studies were used to verify theoretic uncertainty calculations from a simulation program. A swine model was used to evaluate the difficulty and duration of the implant technique, the suitability of the vertebral process as an implant site, vertebral motion due to normal respiration, and the ability to target one vertebra with markers in an adjacent vertebra.
Results: Theoretic accuracy studies confirmed that localization accuracy is a function of marker separation. Phantom studies involving 296 measurements showed that individual implants could be localized within +/-0.25 mm. The largest targeting error observed in 3,600 measurements of 100 implant configurations was 1.17 mm. The implant procedure took 5-10 minutes per site. No significant migration of implants was observed up to 35 days postimplantation, and respiratory motion had no detectable influence on vertebral position. Adjacent vertebrae may be useful for targeting one another with a small sacrifice in localization accuracy.
Conclusions: The use of implanted markers for localization of spinal malignancies has potential for applications in stereotactic radiotherapy. Phantom measurements suggest that localization accuracy similar to intracranial stereotactic radiotherapy techniques is achievable. Swine studies suggest that the implant technique is expedient and feasible for tumor targeting purposes.