Frameless image-guided neurosurgical techniques can achieve high degrees of accuracy when skull-implanted fiducials are used for registration. However, fiducial placement is invasive and uncomfortable for patients. Development of a noninvasive registration method for accurate image-guided functional neurosurgery such as deep brain stimulator placement would therefore be highly desirable. We performed an initial series of experiments using a commercially available fluoroscopic registration package to assess the feasibility of this approach for image-guided functional neurosurgery. We also evaluated the accuracy of landmark placement in the fluoroscopic images using the navigational capability of the software.A fluoroscopic target was created by etching a hexagonal pattern of 1-mm diameter holes on a copper-clad board (0.0254-mm copper cladding on fiberglass). The target was then mounted in a plastic phantom skull, oriented in a mid-sagittal plane. Five implantable fiducial markers were screwed into the phantom in positions which approximated those commonly used clinically. 1.25-mm CT slices were obtained, uploaded to a Stealthstation neuronavigational system and were displayed using the Fluoromerge software package. Lateral and AP images were generated with 2 approximately orthogonal views of the phantom. Registration was carried out both fluoroscopically and using the implanted fiducials. Targets were localized using both methods and the localization errors recorded.Localization error was less than 1 mm using fiducial-based registration, and between 0.8 and 2.9 mm using fluoroscopic registration. Error varied depending on location within the volume of the phantom.Initial experiments show that fluoroscopic registration is feasible for the performance of frameless functional neurosurgical procedures, although accuracy is still insufficient. Intraoperative verification of lead location was also shown to be feasible in one case.
View details for DOI 10.1159/000147635
View details for PubMedID 18663338