Preliminary experience using an optimized three-point transformation algorithm for spatial registration of coordinate systems: a method of noninvasive localization using frame-based stereotactic guidance systems.

This study describes the use of an optimized three-point transformation algorithm to spatially cross-register a volumetric computerized tomographic scan or magnetic resonance image data set with the coordinate system of a stereotactic frame. This algorithm was tested for accuracy using a scanned phantom in which calculated targets, using the Brown-Roberts-Wells (BRW) frame picket-fence algorithm as a standard, could be compared to physical targets measured using a BRW arc and phantom. These target values were then compared to target values calculated with the optimized three-point algorithm. This method was used for target localization in 21 patients. Following this noninvasive localization method, the standard BRW stereotactic system was used for guidance. The application accuracy of this frameless localization technique was within the limits of the scan slice thickness in 16 of 21 cases, with an average error of 2.11 mm in an average scan slice thickness of 3.1 mm. Intracranial targets were successfully reached in all cases without morbidity or mortality. The algorithm can be customized to cross-register image data sets with the coordinate systems of a wide variety of stereotactic guidance systems. The method increases the convenience and flexibility of frame-based stereotactic guidance by providing a means of noninvasive localization that can be accomplished electively at a separate time from the guidance part of a stereotactic operative procedure.