Objective: A custom-designed anthropomorphic head phantom, containing computed tomography (CT) and magnetic resonance (MR) viewable targets, was used in the assessment of stereotactic localization accuracy.
Materials and Methods: The Brown-Roberts-Wells (BRW) or Leksell stereotactic ring was rigidly fixed to the phantom. CT and MR images were then obtained according to radiosurgery protocols with the corresponding localizer frame attached. Plastic spheres and rods appeared at various locations within the phantom, when filled with aqueous solution, and their images served as targets to compute stereotactic target coordinates using software compatible with each frame. Coordinates derived using CT and MR were compared with mechanical measurements obtained using the BRW or Leksell stereotactic arc systems.
Results: For the BRW stereotactic system, the average vector distance to agreement of image-derived coordinates with the mechanical measurements was 1.41 Ψ 0.90 mm (CT) and 1.37 Ψ 0.38 mm (MR). Similar results were obtained using the Leksell system: 0.78 Ψ 0.33 mm (CT) and 1.45 Ψ 0.86 mm (MR). The vector distance to agreement between CT and MR was 1.42 Ψ 0.55 mm for the BRW and 1.31 Ψ 0.60 mm for the Leksell systems.
Conclusions: The data support the use of our anthropomorphic phantom, and present a methodology for assessing radiosurgery target localization and imaging accuracy.