Automated electrocorticographic electrode localization on individually rendered brain surfaces

https://doi.org/10.1016/j.jneumeth.2009.10.005Get rights and content

Abstract

Brain surface electrocorticographic (ECoG) recordings can investigate human brain electrophysiology at the cortical surface with exceptionally high signal to noise ratio and spatio-temporal resolution. To be able to use the high spatial resolution of ECoG for accurate brain function mapping and neurophysiology studies, the exact location of the ECoG electrodes on the brain surface should be known. Several issues complicate robust localization: surgical photographs of the electrode array made after implantation are often incomplete because the grids may be moved underneath the skull, beyond the exposed area. Computed tomography (CT) scans made after implantation will clearly localize electrodes, but the effects of surgical intervention may cause the exposed brain to move away from the skull and assume an unpredictable shape (the so-called brain shift). First, we present a method based on a preoperative magnetic resonance imaging (MRI) coregistered with a post-implantation CT scan to localize the electrodes and that automatically corrects for the brain shift by projecting the electrodes to the surface of the cortex. The calculated electrode positions are visualized on the individual subjects brain surface rendering. Second, the method was validated by comparison with surgical photographs, finding a median difference between photographic and calculated electrode centers-of-mass of only 2.6 mm, across 6 subjects. Third, to illustrate its utility we demonstrate how functional MRI and ECoG findings in the same subject may be directly compared in a simple motor movement experiment even when electrodes are not visible in the craniotomy.

Introduction

One of the latest additions to the cognitive neuroscience toolbox is electrocorticography (ECoG), where detailed information about the regional and functional organization of the brain is obtained from patients who are implanted with cortical electrodes for diagnostic purposes. Human ECoG is unique in the detail of electrical signal properties (e.g. spatial (Cooper et al., 1965) and temporal (Miller et al., 2009) resolution), and is growingly applied to cognitive paradigms in the service of cognitive neuroscience. Although patients, typically suffering from epilepsy, exhibit abnormal activity in some brain regions, most of the electrodes cover healthy brain tissue, allowing for extrapolation of findings in cognitive experiments to the normal population.

ECoG recordings measure the electrical potential from the brain surface, using exposed metal electrodes. ECoG recordings are used to functionally identify different brain areas such as motor (Crone et al., 1998a, Crone et al., 1998b, Miller et al., 2007a, Pfurtscheller et al., 2003), language (Crone et al., 2001a, Crone et al., 2001b, Sinai et al., 2005), auditory (Edwards et al., 2005), and visual cortex (Yoshor et al., 2007), or, for example, to study spontaneous neuronal activity (Nir et al., 2008) and neurophysiology (Canolty et al., 2006). The analyses of ECoG electrode signals are done on individual patients and above all are highly specific to the brain tissue from which signal is sampled (Ball et al., 2009). Electrodes are typically 2.3 mm in diameter and measure virtually no signal from immediately adjacent neural tissue. A major problem faced in ECoG research is to identify exactly where these electrodes are located. Rough estimations are, given the size of electrodes, insufficient for application of ECoG to neuroscientific questions regarding the regional and functional organization of the brain.

Several issues complicate accurate localization of these electrodes. First, matching photographs made of the grid after implantation to an MRI scan (Wellmer et al., 2002), are not sufficient, since neurosurgeons try to minimize the size of the craniotomy and will usually slide electrodes under the skull, away from the exposed area. Second, computed tomography (CT) scans, made after implantation, can localize electrode positions (Noordmans et al., 2001), but the shape of the brain surface is generally changed by the surgical procedure. Leakage of CSF after opening of the dura, the thickness of the implanted material, and the general reaction to surgical intervention, may all cause the exposed brain to move away from the skull and assume an unpredictable shape. This brain shift may cause a significant mismatch that can be more than 1 cm between the CT scan and a magnetic resonance image (MRI) scan obtained preoperatively (Dalal et al., 2008, Hill et al., 2000). Third, post-implantation structural MRI scans would offer a solution to this problem (Schulze-Bonhage et al., 2002, Studholme et al., 2001), but the clinical safety guidelines of many institutions prohibit post-implant MRI scans for the risk of electrode induction heating (Bhavaraju et al., 2002).

Apart from the few studies using post-implant MRI scans, all papers on ECoG that we are aware of use either a match of MRI rendering to photos, hence ignoring the electrodes positioned under the skull and out of view of a camera, or ignore the shift after matching CT to MRI. Several studies projected electrode locations to a standardized brain in Talairach coordinates using a method based upon X-rays (Miller et al., 2007b), but this method suffers from the fact that identified electrode locations cannot be linked to subject-specific gyral anatomy, which can vary greatly from person to person. Dalal et al. (2008) approached the problem using operative photos to visually localize the ECoG electrodes on a reconstructed cortex from a preoperative MRI, and combined this with X-rays to include electrodes not visible in the craniotomy. Their manual registration procedure, however, takes quite long, even with experience, and it has not been established whether their method is reproducible across experimenters, or whether it might also work to localize subtemporal or interhemispheric electrodes (where no part of the array is revealed by the craniotomy, making extrapolation less reliable).

Here we present a new method that uses a preoperative MRI coregistered with a post-implantation CT scan to localize the electrodes, and then automatically corrects for the brain shift by projecting the electrodes to the surface of the cortex. It consists of a MATLAB (The Mathworks, Inc., Natick, MA, USA) based package used in combination with SPM5 software (http://www.fil.ion.ucl.ac.uk/spm/). To validate the accuracy of the projection, the auto-registered electrode locations are compared with operative photographs in six patients. For one additional patient we illustrate the usefulness of this method, by showing that electrodes outside the craniotomy can now be included in, for instance, investigation of the relationship between fMRI activation and ECoG.

Section snippets

Patients

Seven patients were implanted with platinum electrodes (AdTech, Racine, WI, USA) for epilepsy monitoring. Electrodes had a diameter of 2.3 mm exposed (4.0 mm overall) and an inter-electrode distance of 1 cm center-to-center. All patients gave written informed consent, and the study was approved by the ethical committee of the University Medical Centre Utrecht, in accordance with the Declaration of Helsinki 2004.

Technique

Before implantation, structural MRI scans were made on a 1.5 T (patients 2, 3, 6 and 7)

Technique

Electrodes were projected to the surface of the brain in the direction orthogonal to the local surface of the shifted cortex (Fig. 1). Supplemental Fig. S2 shows for one subject that a lateral projection would have led to substantially different results. For each patient, the projection procedure, including up to 128 electrodes, took less than 2 h (including preprocessing of MR and CT scans, for any of three users).

Validation

Distances between electrodes on the photo and the projection are shown in Fig. 3

Discussion

The identification of the exact location of the electrodes is an important issue in ECoG research. This study first presents a method to localize ECoG electrodes on an individual, preoperative MRI scan. The MRI was coregistered with a CT scan made after implantation of the ECoG electrodes. The CT was then used to localize the ECoG electrodes and these electrodes were automatically projected on the cortical surface of the MRI. Second, to validate the method, a comparison between projected

Acknowledgements

The authors thank Peter Gosselaar and Peter van Rijen for implantation of the grids, Frans Leijten and Cyrille Ferrier for their help in acquiring the data and Josien Pluim for her advice on image coregistration. This research is supported by the Dutch Technology Foundation STW, applied science division of NWO and the Technology Program of the Ministry of Economic Affairs, and the University of Utrecht, grant UGT7685. We appreciate the enthusiastic participation of the patient's and staff at

References (33)

  • W.M. Wells et al.

    Multi-modal volume registration by maximization of mutual information

    Med Image Anal

    (1996)
  • T. Ball et al.

    Differential representation of arm movement direction in relation to cortical anatomy and function

    J Neural Eng

    (2009)
  • R.T. Canolty et al.

    High gamma power is phase-locked to theta oscillations in human neocortex

    Science

    (2006)
  • N.E. Crone et al.

    Electrocorticographic gamma activity during word production in spoken and sign language

    Neurology

    (2001)
  • N.E. Crone et al.

    Functional mapping of human sensorimotor cortex with electrocorticographic spectral analysis. II. Event-related synchronization in the gamma band

    Brain

    (1998)
  • N.E. Crone et al.

    Functional mapping of human sensorimotor cortex with electrocorticographic spectral analysis. I. Alpha and beta event-related desynchronization

    Brain

    (1998)
  • Cited by (0)

    View full text