Biomedical technologies in neuroprosthetics are made to restore or replace functions from the nervous system, as well as alleviate pathological conditions. Peripheral nerve interfaces use neuromodulatory techniques to inhibit pain or depression, prevent epileptic strokes, or connect amputee patients to an artificial limb. Man-made neural interfaces are, however, far from matching the complexity of the nerves, leaving patients with low-resolution interfaces.
To engineer peripheral neuroprostheses, scientists and clinicians investigate neuromodulatory strategies using in vitro neuron culture, explanted nerves, in vivo and in silico experiments using simulation, patch-clamp or extracellular recordings, and neural classification/learning algorithms. However, the available models can not replicate the complex morphology of the nerves, thereby limiting the resolution and selective recording of axonal activity. While nerve fibers respond differently to a neuromodulation paradigm depending on their size and degree of myelination, studies often fail to deal with the whole variety of axon characteristics present in the body.
Therefore, in this thesis, I developed a nerve-on-a-chip platform that uniquely integrates the following features: stimulation and recording of single myelinated nerve fibers, controlled single-fiber action potential propagation with respect to electrodes position, and signal amp- lification leading to recorded signals in the 20 - 300 ÎŒV range. Nerve fibers diameter can further be deduced from propagation velocity to include the effect of nerve heterogeneity and function in the implementation of neuromodulatory strategies.
I first present a versatile nerve-on-a-chip platform, aiming at offering robust standards to eval- uate neuromodulation strategies and selectivity of neuroprostheses. In order to conduct rapid, high resolution and systematic recording of peripheral nerve activity, microchannel electrodes are used to achieve stimulation and recording of single myelinated nerve fibers ex vivo. I demonstrate the platform versatility through three case studies: (i) Optimising the design of microchannel regenerative implants to detect the velocity and direction of neural signal propagation. I implemented and experimentally validated a mathematical model predicting how the reliability of velocity calculation depends on the implant geometry and neural signal waveform. (ii) Demonstrating how heat-induced neuroinhibition upon P3HT:PCBM (blend polymer) illumination selectively and reversibly silenced thinner fibers, but did not affect larger fibers. (iii) To improve artificial sensory feedback, new patterns of electrical stimulation were investigated in silico. In this study, the platform is used to show that high frequency modulated stimulation elicits more biomimetic nerve activity. This novel nerve-on-a-chip platform enables rapid prototyping using standard microfabrication methods, simultaneous probing of myelinated fibers of various diameters, data library collection for computational modelling and multi-factorial analysis.
Subsequently, I have applied the results from (i) to develop a microfabrication process of a microchannel regenerative implant, further exploiting the use of microchannel electrodes to improve neural interfaces.
The nerve-on-a-chip platform is an efficient tool for designing and evaluating new implantable neuroprosthetic devices, using time and cost-efficient techniques. We foresee that this simple and broad
EPFL2019
