Abstracts

Rationale: Many epilepsy centers use intracranial electroencephalograpy (iEEG) to localize seizures and increasingly for studies of human cognition in patient volunteers. Interpreting iEEG results requires knowledge of the relative position and anatomic location of implanted electrodes. This can be challenging based on cross sectional or static imaging alone, particularly for complex implantations with multiple stereotactic electrodes, overlapping strips and grids, and variable naming conventions. To facilitate interpretation and reduce errors among the typical interdisciplinary team of neurosurgeons, epileptologists, radiologists and researchers, we sought to build a cost effective web-based 3D visualization for iEEG. Methods: Inputs to the visualization are an individual patient's cerebral cortical surfaces, reconstructed with Freesurfer (freesurfer.net) based on a pre-implantation volumetric whole brain T1-weighted MRI scan, and surface and/or depth electrode contact coordinates derived from a co-registered post-implantation CT. The set of cortical surfaces and electrodes are rendered in Blender (blender.org), free and open-source software for 3D modeling and animation, in automated fashion using a Python script. To construct the interactive 3D web application, we use Blend4Web (blend4web.com), a free and open-source framework for displaying and manipulating Blender scenes in web browsers using the Web Graphics Library (WebGL) JavaScript standard. Results: Combining different free and open source components, we produced interactive web-based 3D renderings of intracranial electrodes that complement other available software for data visualization. Additional custom code allows users to show or hide model components or add transparency or colors to cortical regions. Electrode contacts may be modified with colors and labels, and these annotations can be animated to capture seizure propagation and written to or retrieved from a local file or remote database. Hovering over cortical regions displays the corresponding anatomic labels. Functional imaging results can also be displayed on individual cortical surfaces. The 3D view is integrated with multiplanar cross sectional imaging, so hovering over an electrode reveals contact names and annotations as well as their corresponding locations on fused MRI-CT images in separate panels. This combination of linked 3D and planar views helps to find contacts of interest quickly and aids in understanding lesion proximity and seizure spread. Conclusions: Our web-based 3D visualization for iEEG leverages the power of modern devices, web browsers and software standards. Users only need a url to load and share the interactive visualization, which runs like any other website on personal computers, tablets and smartphones. This approach can facilitate collaboration and communication across research and/or clinical teams at one or more institutions. Alone or embedded in another website, such visualizations may also be useful for patient education. Shared in an online repository, the code base for our application can be further customized and extended by other researchers and clinicians. Funding: This work was supported by the DARPA Restoring Active Memory (RAM) program (Cooperative Agreement N66001-14-2-4032). The views, opinions, and findings expressed are those of the authors and should not be interpreted as representing the official views or policies of the Department of Defense or the U.S. Government.