Abstracts

Automated Invasive Electrode Labeling and Location Validation

Abstract number : 1.155
Submission category : 3. Neurophysiology / 3C. Other Clinical EEG
Year : 2018
Submission ID : 501002
Source : www.aesnet.org
Presentation date : 12/1/2018 6:00:00 PM
Published date : Nov 5, 2018, 18:00 PM

Authors :
Kenneth Taylor, Cleveland Clinic; John C. Mosher, Cleveland Clinic; Dileep R. Nair, Cleveland Clinic; Richard M. Leahy, University of Southern California; and Anand Joshi, University of Southern California

Rationale: Stereoelectroencephalography (SEEG) is an invasive surgical procedure that is used to identify the areas of the brain which are involved in epileptic seizures. Electrodes placed in targeted areas of the brain are then monitored in order to observe ictal activity. Here we describe an automated approach to contact labeling which uses the subject’s MRI and CT to register and transfer anatomic labels from a common atlas to the SEEG data, which can then be visualized to provide an informative display of the implantation. Methods: We demonstrate the method for labeling and visualizing SEEG implantations using Brainstorm (Tadel et al. 2011). First, the subject MRI is segmented into 3D regions and labeled, here using Brainsuite (Shattuck et al. 2002). Anatomical parcellation of the brain is according to the BCI-DNI atlas (http://brainsuite.org/svreg_atlas_description/). A functional subparcellation of these regions is then utilized as described in the US Brain atlas (Joshi et al. 2017). Each region is then assigned a color so as to be easily distinguishable from adjacent regions. Separately, the CT is co-registered to the MRI using MATLAB (The Mathworks Inc.) in order to align the implanted electrodes with the 3D reconstruction of the brain. This allows us to observe the region in which each contact is located. In order to simplify the visualization, we represent each contact by a sphere, and then color the spheres according to their location. To address contacts on a boundary, the spheres are segmented to allow multiple colors. Results: Fig.1 shows the subparcellation of the brain as determined by the USC Brain atlas, a full 3-D labeling of the brain volume. Fig. 2 shows how this visualization can be combined with a display of the contact locations in order to create a more informative display. On the left, the SEEG electrodes are shown as registered and viewed to the patient's transparent cortical surface. In the center, the patient's full 3D labeling as registered and transferred from the Atlas. On the right, the contacts from each SEEG electrode, each now a sphere colored by its labelled region. In order to validate the automatically assigned labels, a comparison is made against those assigned by clinical fellows during case review. The labels, particularly in cases of disagreement between the two sets, are reviewed by an expert to determine the correct label in each case. Conclusions: The methods described herein allow us to visualize SEEG implantations using color to represent contact locations relative to region boundaries. The automated labeling process can also be used to generate tables summarizing implantation locations as well as automatically order presentations of data using anatomical location. Funding: A Brain Atlas for Mapping Connectivity in Focal Epilepsy NIH 2018 R01