Intraoperative Stereotactic Frame Registration Using a Three-Dimensional Imaging System with and without Preoperative Computed Tomography for Image Fusion

Abstract

Background: The O-arm O2 imaging system (OAO2) is an intraoperative cone beam 3D tomogram imaging tool with a wide enough field of view to perform intraoperative fiducial registration with standard stereotactic frames. However, the OAO2 3D images (cone beam CT) provide limited tissue contrast, which may reduce the accuracy of fusion to a preoperative targeting MRI for planning awake deep brain stimulation (DBS) surgeries. Therefore, most users obtain a preoperative CT scan to use as the reference exam for computational fusion with the preoperative targeting MRI and the intraoperative OAO2 cone beam CT. Objective: In this study, we retrospectively analyzed the discrepancy between stereotactic coordinates of deep brain targets on MRI derived from intraoperative OAO2 fiducial registration with and without the use of preoperative CT as the reference for image fusion. Methods: Preoperative stereotactic CT/MRI and intraoperative OAO2 cone beam CT were retrospectively evaluated for 27 consecutive DBS patients, using two commercial surgical planning software packages (BrainLab Elements and Medtronic Stealth 8). The anterior commissure, posterior commissure, and left subthalamic nucleus were identified on preoperative MRI. Each patient had intraoperative fiducial registration using the OAO2 with a Leksell headframe. For each subject, the reference scan for image fusion was set as either the preoperative CT or the preoperative MRI (volumetric T1 with contrast). Computed stereotactic coordinates for each target were then compared. Results: For 8 of 27 subjects, a discrepancy greater than 1.0 mm for at least one designated target was observed utilizing the Medtronic Stealth S8 planning station when a preoperative CT scan was not used. An additional 5 (5/27) had a discrepancy greater than 2 mm. The most common discrepancy was in the z axis. No coordinate discrepancies greater than 1 mm were observed utilizing BrainLab Elements. Conclusions: Caution is advised in fusing intraoperative OAO2 images directly to preoperative MRI without a preoperative CT as the reference exam for image fusion, as the specific fusion algorithm employed may unpredictably affect targeting accuracy.