Three-dimensional multilayer concentric bipolar electrodes restrict spatial activation in optic nerve stimulation

Objective. Intraneural nerve interfaces often operate in a monopolar configuration with a common and distant ground electrode. This configuration leads to a wide spreading of the electric field. Therefore, this approach is suboptimal for intraneural nerve interfaces when selective stimulation is required. Approach. We designed a multilayer electrode array embedding three-dimensional concentric bipolar (CB) electrodes. First, we validated the higher stimulation selectivity of this new electrode array compared to classical monopolar stimulation using simulations. Next, we compared them in-vivo by intraneural stimulation of the rabbit optic nerve and recording evoked potentials in the primary visual cortex. Main results. Simulations showed that three-dimensional CB electrodes provide a high localisation of the electric field in the tissue so that electrodes are electrically independent even for high electrode density. Experiments in-vivo highlighted that this configuration restricts spatial activation in the visual cortex due to the fewer fibres activated by the electric stimulus in the nerve. Significance. Highly focused electric stimulation is crucial to achieving high selectivity in fibre activation. The multilayer array embedding three-dimensional CB electrodes improves selectivity in optic nerve stimulation. This approach is suitable for other neural applications, including bioelectronic medicine.

Fabrication and validation of an intraneural interface tailored for optic nerve stimulation

Eleonora Borda

Technological progress in materials science and microengineering along with new discoveries in neuroscience have contributed to restore lost or impaired sensory functions by closely interfacing with the nervous system. Electronic devices have begun to be integrated within the body to artificially reproduce or record the electrical encoding of neural signals, the language of the nervous system. This has allowed us to restore the impaired communication between the missing or defective organ and the brain. As evidence of this, cochlear implants have returned hearing to millions of people world wide. On the other hand, there is no suitable solution yet for the 43 millions people who are currently blind. Nevertheless, several research groups and companies are trying to tackle this issue. So far, different devices have demonstrated the capability of restoring a certain degree of artificial vision, however, with limited success in delivering meaningful improvements in daily life activities. This is because, unlike hearing, vision is an extremely complex sense and requires higher stimulation resolution. Limitations of retinal prostheses have pushed further the investigation of other targets along the visual pathway, such as the optic nerve, the thalamus and the visual cortex. Among these, our group has started to explore the possibility of stimulating the optic nerve. The main advantages are the direct stimulation of the axon fibers and the access to the entire visual field with only one device. It is also interesting because it could work for those patients who are not eligible for retinal prostheses, due to severe eye trauma or retinal detachment, and it does not require optical transparency.In this thesis, we investigate stimulation strategies and materials to improve the outcome of optic nerve prostheses and make them one step closer to a viable option for blind patients. We first propose the use of machine learning algorithms to classify cortical activity upon electrical stimulation of the optic nerve. Secondly, we investigate a strategy to achieve higher spatial selectivity of the optic nerve stimulation. We design a three-dimensional (3D) multilayer concentric bipolar (CB) configuration and fabricate a modified version of the OpticSELINE. We validate the efficacy of the novel design during in-vivo experiments on rabbits and find that cortical activation is reduced when using the 3D multilayer CB electrodes compared to the classic monopolar configuration. In parallel, we explore the use of novel polymers, namely Off-Stoichiometry Thiol- Ene Epoxy (OSTE+), to design conformable neural interfaces. Their chemistry makes them a suitable candidate for easy integration in cleanroom processes and they show superior adhesion properties with respect to standard thin films currently in use. Hence, we characterise their usability as substrate and encapsulation for neural interfaces in-vitro and in-vivo. Lastly, we propose an alternative fabrication technique, ink-jet printing of platinum on flexible substrates, for neurophysiology applications. Altogether, the findings highlighted in this thesis will hopefully lead to a more in-depth investigation of the optic nerve prosthesis and can also be relevant for other applications in the field of neuroprosthetics.

EPFL2022

Spatially selective activation of the visual cortex via intraneural stimulation of the optic nerve

Fiorenzo Artoni, Vivien Gaillet, Diego Ghezzi, Dario Lipucci Di Paola, Silvestro Micera, Ariastity Mega Pratiwi, Paola Vagni

Retinal prostheses can restore a functional form of vision in patients affected by dystrophies of the outer retinal layer. Beyond clinical utility, prostheses for the stimulation of the optic nerve, the visual thalamus or the visual cortex could also serve as tools for studying the visual system. Optic-nerve stimulation is particularly promising because it directly activates nerve fibres, takes advantage of the high-level information processing occurring downstream in the visual pathway, does not require optical transparency and could be effective in cases of eye trauma. Here we show, in anaesthetized rabbits and with support from numerical modelling, that an intraneural electrode array with high mechanical stability placed in the intracranial segment of the optic nerve induces, on electrical stimulation, selective activation patterns in the visual cortex. These patterns are measured as electrically evoked cortical potentials via an ECoG array placed in the contralateral cortex. The intraneural electrode array should enable further investigations of the effects of electrical stimulation in the visual system and could be further developed as a visual prosthesis for blind patients.

2019

Biologically plausible unsupervised learning in shallow and deep neural networks

Bernd Albert Illing

The way our brain learns to disentangle complex signals into unambiguous concepts is fascinating but remains largely unknown. There is evidence, however, that hierarchical neural representations play a key role in the cortex. This thesis investigates biologically plausible models of unsupervised learning of hierarchical representations as found in the brain and modern computer vision models. We use computational modeling to address three main questions at the intersection of artificial intelligence (AI) and computational neuroscience.The first question is: What are useful neural representations and when are deep hierarchical representations needed? We approach this point with a systematic study of biologically plausible unsupervised feature learning in a shallow 2-layer networks on digit (MNIST) and object (CIFAR10) classification. Surprisingly, random features support high performance, especially for large hidden layers. When combined with localized receptive fields, random feature networks approach the performance of supervised backpropagation on MNIST, but not on CIFAR10. We suggest that future models of biologically plausible learning should outperform such random feature benchmarks on MNIST, or that such models should be evaluated in different ways.The second question is: How can hierarchical representations be learned with mechanisms supported by neuroscientific evidence? We cover this question by proposing a unifying Hebbian model, inspired by common models of V1 simple and complex cells based on unsupervised sparse coding and temporal invariance learning. In shallow 2-layer networks, our model reproduces learning of simple and complex cell receptive fields, as found in V1. In deeper networks, we stack multiple layers of Hebbian learning but find that it does not yield hierarchical representations of increasing usefulness. From this, we hypothesise that standard Hebbian rules are too constrained to build increasingly useful representations, as observed in higher areas of the visual cortex or deep artificial neural networks.The third question is: Can AI inspire learning models that build deep representations and are still biologically plausible? We address this question by proposing a learning rule that takes inspiration from neuroscience and recent advances in self-supervised deep learning. The proposed rule is Hebbian, i.e. only depends on pre- and post-synaptic neuronal activity, but includes additional local factors, namely predictive dendritic input and widely broadcasted modulation factors. Algorithmically, this rule applies self-supervised contrastive predictive learning to a causal, biological setting using saccades. We find that networks trained with this generalised Hebbian rule build deep hierarchical representations of images, speech and video.We see our modeling as a potential starting point for both, new hypotheses, that can be tested experimentally, and novel AI models that could benefit from added biological realism.