Automated registration of intraoperative CT image data for navigated skull base surgery.

OBJECTIVES: With a new intraoperative computed tomography (CT) imaging system, patient-to-image registration without any invasive registration markers is possible. Furthermore, registration can be performed fully automatically. The accuracy of this method for skull base surgery was investigated in this study. METHODS: We employed a phantom study design. A phantom skull was equipped with 33 target markers in the regions of the anterior and lateral skull base. CT image data were acquired with an intraoperative CT suite. Image data were transferred as DICOM data to the navigation system, and registration was performed automatically. For registration, the position of the patient and the position of the CT gantry were monitored in the imaging process, using the infrared camera of a navigation system. Using the pointing device of the navigation system, the target markers were identified. The accuracy was measured as the spatial difference of the target markers in image space and on the phantom. RESULTS: Accuracy was always sufficient for image-guided surgery of any region of the skull base, with an average target registration error of below 1.2 mm. In contrast to traditional non-invasive registration methods, there was no difference in registration accuracy between the anterior skull base and the lateral skull base. CONCLUSIONS: Fully automated registration based on a tracked CT gantry is a robust and accurate registration method for skull base surgery.