PURPOSEThis study was undertaken to determine the impact of geometric distortions on the spatial accuracy of magnetic resonance imaging (MRI)-guided stereotactic localization for gamma knife functional radiosurgery.METHODThe spatial accuracy of MRI was evaluated by comparing stereotactic coordinates of intracranial targets, external fiducials, and anatomic structures defined by computed tomographic and MRI studies of the Radionics skull phantom (Radionics, Inc., Burlington, MA), the Rando head phantom, and 11 patients who underwent gamma knife functional radiosurgery. The distortion in MRI was assessed from computed tomographic and MRI fusion studies for these patients, as well as from MRI studies acquired by swapping the direction of the magnetic field gradients for five patients who underwent gamma knife radiosurgery and three patients who underwent MRI-guided frameless surgery. A follow-up program to compare the location of the created lesion with the intended target complemented the analysis.RESULTSThe average difference between computed tomographic and MRI stereotactic coordinates of external fiducials, intracranial targets, and anatomic landmarks was of the order of 1 pixel size (0.9 × 0.9 × 1 mm3) along thex,y, andzaxes. The average linear scaling along these axes as determined by fusion studies was approximately 0.8% and consistent with a single pixel. The follow-up studies, available for seven patients, revealed good agreement between the location of the created lesion and the intended target.CONCLUSIONThe spatial accuracy of an MRI-based localization system can be comparable to computed tomography-based localization with the added benefit of MRI resolution. Both machine- and object-related MRI distortions can be reduced to an acceptable level with contemporary scanners, optimized scanning sequences, and distortion-resistant stereotactic instruments.