Accuracy of stereotactic localisation using magnetic resonance imaging: a comparison between two- and three-dimensional studies.

The accuracy of stereotactic localisation using magnetic resonance (MR) imaging has been assessed in a phantom study. Parallel studies compared the accuracy obtained. First, a series of two-dimensional (2D) MR slices (transverse, coronal, and sagittal) was acquired sequentially to image the three-dimensional (3D) volume of the phantom. Then, the same volume was imaged in a 3D MR study in which the entire volume was excited simultaneously and 2D slices in transverse, coronal, and sagittal planes were then reconstructed from the 3D data set. The results showed that the 3D acquisition gave superior results in all three planes, and overall it was found that only 1% of the phantom volume was affected by an error greater than 2 mm, compared with 11% for the 2D study. New facilities for image quality assurance provided in GammaPlan version 3.0 have been tested against images from the 2D study with known distortion and consequent localisation errors and successfully identified all images in which the localisation errors were likely to be greater than 3.7 mm.