Introduction: Intracranial electrode implantations for DBS and seizure localization are common functional neurosurgical procedures. In order to refine therapeutic targets and draw meaningful conclusions from electrophysiological data, intracranial electrodes need to be accurately localized. The demand for simpler and more efficient methods to localize implanted electrodes has grown as the volume of these procedures continues to increase. We developed a semi-automated pipeline that integrates pre-operative MRI and post-operative CT data to determine electrode locations.
Methods: The semi-automated pipeline was tested with a sample of patients (n=9) who underwent implantation of sEEG, surface grid/strip, or a combination of these electrode configurations. We co-registered the pre-op MRI with the post-op CT (3D Slicer, Boston, MA) to align electrodes in patient-specific anatomy and extracted a volumetric brain image (FSL, Oxford, UK) to create a mask that represented the dataspace of interest (MATLAB, Natick, MA). The image was processed using size and intensity thresholding, erosion, and Gaussian kernel convolution. We determined coordinates for electrodes’ centroids, which allowed for sophisticated visualization. Coordinate accuracy was evaluated by comparison to coordinates generated from a validated manual method for electrode localization.
Results: Coordinates for approximately 600 contacts on the implanted electrode were successfully computed for all patients. Visualization demonstrated that automation did not affect the number of electrodes detected. Coordinates were compared to output from a validated manual method for electrode localization, and accuracy was non-inferior.
Conclusions: The method described is an accurate and easily-implemented method for intracranial electrode localization using MATLAB and open access software. Compared to similar open access methods, our pipeline requires minimal user input, which significantly reduces person-hours required for task completion. From a clinical perspective, this pipeline allows for seamless retrospective analysis of surgical targets and thus has the potential to inform prospective image-guided surgical protocols.
Patient Care: As mentioned in the introduction, the ability to accurately and efficiently define precise locations of intracranial electrodes is essential for refining therapeutic targets for stimulation. On a more longitudinal scale, once fully realized, this technique could also be employed for neuroelectrophysiology research to elucidate mechanisms of important cognitive processes, such as memory and mood regulation, that are impaired in pathological states and could inform future therapeutic interventions.
Learning Objectives: By the conclusion of this session, participants should be able to: 1) Describe the importance of computer-based image processing for advancing neurological electrophysiology research and refining therapeutic targets for functional surgery, 2) Discuss how the use of thresholding and convolution with a gaussian kernel can be used to isolate individual electrode leads within an image, 3) Identify the limitations of current image processing techniques, including the absence of a fully automated process for image coregistration and electrode localization with three-dimensional visualization.