Simultaneous recording of ECoG and intracortical neuronal activity using a flexible multichannel electrode-mesh in visual cortex

Abstract

Electrocorticogram (ECoG) is a well-balanced methodology for stably mapping brain surface local field potentials (LFPs) over a wide cortical region with high signal fidelity and minimal invasiveness to the brain tissue. To directly compare surface ECoG signals with intracortical neuronal activity immediately underneath, we fabricated a flexible multichannel electrode array with mesh-form structure using micro-electro-mechanical systems. A Parylene-C-based “electrode-mesh” for rats contained a 6 × 6 gold electrode array with 1-mm interval. Specifically, the probe had 800 × 800 μm² fenestrae in interelectrode spaces, through which simultaneous penetration of microelectrode was capable. This electrode-mesh was placed acutely or chronically on the dural/pial surface of the visual cortex of Long–Evans rats for up to 2 weeks. We obtained reliable trial-wise profiles of visually evoked ECoG signals through individual eye stimulation. Visually evoked ECoG signals from the electrode-mesh exhibited as well or larger signal amplitudes as intracortical LFPs and less across-trial variability than conventional silver-ball ECoG. Ocular selectivity of ECoG responses was correlated with that of intracortical spike/LFP activities. Moreover, single-trial ECoG signals carried sufficient information for predicting the stimulated eye with a correct performance approaching 90%, and the decoding was significantly generalized across sessions over 6 hours. Electrode impedance or signal quality did not obviously deteriorate for 2 weeks following implantation. These findings open up a methodology to directly explore ECoG signals with reference to intracortical neuronal sources and would provide a key to developing minimally invasive next-generation brain–machine interfaces.

Introduction

The local field potential (LFP) is a particular class of electrophysiological signals related to a summation of neural activity that has recently become of converging interest (Fries, 2009). Whereas the action potential or spike activity represents output signals from neurons typically located within 100 micrometers of an electrode tip, LFP is considered to arise primarily from excitatory and inhibitory dendritic potentials and afterdepolarization/hyperpolarization (Mitzdorf, 1985) and thus serve as a marker of inputs to and local processing within a volume of brain tissue extending at least several hundred micrometers (Katzner et al., 2009). LFP is also closely correlated with BOLD fMRI signals (Logothetis et al., 2001). Since various frequency components of LFPs have been implicated in carrying across cortical areas relevant sensory, motor or cognitive information (Buschman & Miller, 2007, Whittingstall & Logothetis, 2009), the need for recording LFPs from the global cortical circuits is increasing.

Electrocorticogram (ECoG) is one of the methodologies that enables global mapping of LFPs both in humans and animal models. Compared with microelectrode methods, ECoG has the advantage to record cortical surface LFPs less invasively across broader regions. Compared with scalp electroencephalogram, ECoG has better spatial resolution and signal fidelity (Chao et al., 2010, Gaillard et al., 2009, Leuthardt et al., 2006, Slutzky et al., 2010). In particular, high-density ECoG electrode array is a promising tool for electrophysiological mapping with millimeter precisions for source localization (Hollenberg et al., 2006, Rubehn et al., 2009, Yeager et al., 2008). However, it has been technically demanding to directly compare between surface ECoG and intracortical activity just underneath, because (1) penetrating the ECoG sheet is virtually impossible without damaging the microelectrode tip, and (2) the recording surface of ECoG electrode is too large (typically more than 2 mm in diameter) to allow simultaneous depth-recording within submillimeter.

To address this issue, we developed a novel flexible ECoG probe (electrode-mesh) using micro-electro-mechanical systems (MEMS) technology (Takeuchi et al., 2005). The probe was designed to have a mesh structure for stable electrode contact to the curved brain surface, and for simultaneous penetration of microelectrodes through the fenestrae, or mesh holes. To test the feasibility of our method for combined ECoG and intracortical recordings, we applied the probe in vivo to the primary visual cortex (V1) of Long–Evans rats in acute as well as chronic preparation. We tested if multichannel ECoG signals could be reliably recorded in response to visual stimulation and examined correlation between epidural ECoG and intracortical spikes in terms of the ocular dominance index. We also evaluated the potential advantage of our method over conventional approaches by directly comparing the amplitude of visually evoked potentials or trial-to-trial signal variability among different probes.

Flexible ECoG probe is not only a promising electrophysiological tool for characterizing ensemble neural activity to code particular sensory stimuli or motor plans, but also an important candidate for the input device of brain–machine interfaces (BMIs) that predict stimuli or behaviors from neural responses (Kamitani & Tong, 2005, Lebedev & Nicolelis, 2006). Since safety of ECoG has been clinically ensured (Leuthardt et al., 2006), the ability to decode sensory- or motor-associated information from trial-wise ECoG signals would be critical for development of efficient BMI (Schwartz et al., 2006). Therefore, we tested whether single-trial ECoG signals contained sufficient information for predicting the stimulated eye by decoding analysis (Kamitani and Tong, 2005).