Carbon-based neural interfaces to probe retinal and cortical circuits with functional ultrasound imaging in vivo

Overview & Synopsis

Brain-computer interfaces (BCIs) based on electrode arrays can either stimulate/write information to neurons through electrical current injection, record/read from neurons. Such BCIs have created great hopes in patients with neurological diseases or handicaps but current interfaces display large metallic electrodes facing tissue inflammation leading to gliotic and fibrotic reactions, which in the end limits the device efficacy. In contrast, carbon-based electrodes gained interest because of their combined mechanical, electrical and electrochemical properties. My project focuses on this question: can we use carbon-based electrodes for reading and/or writing efficiently information to neurons? In collaboration with 3 research teams across Europe, this doctoral work explores three different BCIs for different neural applications, with the help of functional ultrasound imaging (fUS) to monitor activity in cortical and subcortical regions. This brain imaging technique relies on the measure of cerebral blood flow (CBF) which correlates with neural activity through neurovascular coupling. In the scope of NEURODIAM (project #1), a European Union (EU) Horizon 2020 funded project, I worked on the proof-of-concept of a full-diamond electrocorticography array (ECoG), combining the insulating properties of intrinsic diamond as passivation layer and the high conductivity of boron-doped diamond for neural interfacing. Upon in vitro characterization of the electrodes, I recorded visual evoked potentials (VEPs) using such implants during light flashes on anesthetized rats and mice. We report impedances in the range of the tens of kΩ and an electrochemical potential window of 3 V for electrodes of 200, 400 or 600 µm of diameter. Observation of the classical shape of VEPs and the reasonable signal-to-noise ratio give us confidence in the functionality of such electrodes. In the scope of the EU-funded Graphene Flagship Core 3 (project #2), I used simultaneously graphene-solution-gated field-effect transistors (gFETs) and fUS to record cortical spreading depolarizations (CSDs) and seizures in acute models of epileptic rats. During CSDs, cerebral blood volume (CBV) spreads to subcortical areas such as hippocampus, in addition to cortical regions. We also observe a tight phase coupling during seizures between infra-slow brain signals and CBV but also oscillations of CBV prior to seizures onset. This new technological platform can help to investigate how both fast and infraslow brain signals contribute to NVC in neurological diseases. In its reduced graphene oxide (rGO) form, graphene has a high charge injection capacity and stability, hence a potential application to retinal prostheses for visually impaired patients (project#3). Our biocompatibility study highlighted no difference in microglia number between controls and rGO-implanted retinas, suggesting a non-inflammatory reaction. In vivo subretinal electrical stimulation of the implant with 10 ms pulses activated the visual structures, observed thanks to fUS. Spatial extent of the activation in the superior colliculus was dependent on the electrode size only at high current amplitudes. Our results suggest that rGO is a reliable material for neural stimulation but can also be applied for neural recording in bimodal prosthetics. In summary, I worked on three different projects involving different carbon-based interfaces for either reading or writing neural activity in vivo. I first established the proof-of-concept of a full-diamond ECoG for neural recording. I then focused on the multimodal recording and analysis CSDs and seizures using gFETs and fUS. Finally, I confirmed the in vivo long-term biocompatibility and functionality of rGO electrodes for subretinal prostheses. All together, these results provide great confidence in carbon-based technology for translation in clinical applications and for the elaboration of future-generation BCIs.