Abstracts

Rationale: Optimal electrode coverage is critical for the success of the intracranial EEG recordings so that limited spatial sampling does not yield an incorrect localization of the epileptogenic zone or false negative results. Multimodal imaging available on the neuronavigation system at the time of implantation surgery will provide additional important landmarks which can assist in more accurate and safe placement of electrodes according to the implantation plan. We have developed a method for fusion of multimodal imaging information for presurgical planning and import of this information on the neuronavigation system for use at the time of surgery.Methods: Images of different modalities (MRI, SPECT, PET, fMRI, DTI, EEG-fMRI, MEG) carrying useful landmarks to plan the implantation of intracranial electrodes are all individually coregistered to the structural MR image which will be used for navigation during surgery. The coregistration techniques are chosen to account for distortions specific to each neuroimaging modality so that the discordance with real anatomy is minimal. Selected landmarks (areas showing e.g. eloquent cortex responsible for language or hand/foot movements, white matter tracts, blood vessels, etc) are segmented and fused into the navigation MR image. The resulted image is used to create 3D surface models within the neuronavigation system. The 3D models of different landmarks are color coded to help easier differentiation and avoid confusion during surgery. The accuracy of localisation (using blood vessels as optimal landmark) was estimated intraoperatively using neuronavigation system. Results: The method has been applied during surgery in six patients with refractory epilepsy. The results for two patients are shown on the Figure: (a, c) intraoperative images as available on the Stealth neuronavigation system (Medtronic, Inc); (b, d) co-registered landmarks along with post-operative reconstruction of the intracranial electrodes. Color coding: Figure b red lesion, pink hyper metabolism on ictal SPECT, dark green hand area, dark blue cortico-spinal tract, orange language areas, light green spike-related activation, light blue veins; Figure d red lesion, dark blue optic radiation tract. High accuracy of placement of 3D models has been confirmed intraoperatively for blood vessels in the vicinity of seizure onset zone. Conclusions: Availability of the 3D models of various relevant landmarks provides a significant qualitative extension of neuronavigation tools, aiding accurate and safe placement of intracranial electrodes. The key advantage of our method is its independence of image analysis tools implemented in neuronavigation system. This allows immediate implementation of the most recent achievements in neuroimaging for use during neurosurgery. The reported results show feasibility and reliability of our method.