Fast nonlinear registration applied to stereotactic functional neurosurgery

Movement disorder surgery depends on patient-specific, intraoperatively acquired neurophysiologic data to determine a final target within a region visualized using preoperative tomographic imaging (MRI or CT). It is desirable to perform inter-individual comparisons on physiologic data for analysis and surgical planning, but they cannot be directly transferred between patients due to intersubject structural variation. A nonlinear transformation that has variable displacement throughout the image volume addresses the small-scale inter-individual differences that global transformations cannot, but computational time requirements prevent its clinical application.;This thesis examines, reviews and provides novel approaches to the problem of intersubject registration of physiologic data in three areas: (1) Accurate MRI to patient registration; (2) fast intersubject nonlinear registration and (3) a software architecture to integrate these components into a surgery analysis and planning system.;Stereotactic headframe registration stabilizes the patient's head and the neurosurgical instruments. An algorithm is presented that recognizes frame fiducials and computes an MRI to frame-space transformation. Tested on 300 256 x 256 x 248 pixel MRI acquisitions, it executes in 0.5 seconds (1 GHz PIII), has a target localization error of 0.5mm and provides MRI quality control by comparing extracted fiducials with their known dimensions.;A multi-threaded nonlinear intersubject registration algorithm is presented that registers two 256 x 256 x 248 voxel T1-weighted MRI acquisitions in 7 minutes (2 x 1GHz PIII) with 0.5mm precision. This algorithm has been validated using test images recovering known applied deformations and intersubject anatomic landmark registration.;Finally, an extensible software framework is presented that integrates the algorithm implementations into a surgical analysis/planning system that has already seen initial clinical exposure in the operating room.