Periodic Reporting for period 1 - HERMES (Hybrid Enhanced Regenerative Medicine Systems)
Reporting period: 2019-01-01 to 2019-12-31
The overall goal of HERMES is to provide the proof-of-concept that a restorative dialogue between bioengineered and mammalian brain tissue can be established and controlled, so to heal brain damage effectively and safely. To overcome the limitations of regenerative medicine and neuroprosthetics, HERMES fosters the emergence of the novel field of enhanced regenerative medicine. This paradigm is rooted in the establishment of biohybrid neuronics (neural electronics), in which the symbiotic integration of bioengineered brain tissue, neuromorphic microelectronics and artificial intelligence (AI) effectively pursues a controlled self-repair process of dysfunctional brain circuits.
In pursuing the proof-of-concept, we will focus on temporal lobe epilepsy (TLE), the most common epileptic syndrome and the most frequently unresponsive to medications. As the hippocampus is primarily stricken in TLE, we aim at replacing its damaged part by bioengineered hippocampal tissue grafting.
Along its roadmap, HERMES also promotes interdisciplinary cross-fertilization beyond the consortium; it extends the concepts of enhanced brain regeneration to philosophy, ethics, policy and society, ultimately fostering the establishment of highly versatile intelligent biohybrids that will potentially be compliant to adapt to cure any brain disorder beyond epilepsy.
For the neuromorphic computing system (NCS) guiding the functional graft-host integration, we have designed and fabricated the CMOS-neuron chip along with the analogue front-end (AFE) enabling a large input signal dynamic range, and we have characterised the electrical properties of memristors of different materials. We have also programmed a Spiking Neural Network in the Digital Signal Processor of a commercial microelectrode array (MEA) electrophysiology system to attain high flexibility in testing and refining the NCS design.
Aiming at the in vivo setting, we have modelled the response of microprobes of different material, shape and size to bending stress, coupling the study with cytotoxicity assays according to the ISO 10993 recommendations. With this, we have identified polyimide as substrate and poly(-ethyl acylate) as encapsulation material and we have selected the initial microprobe design specifications. We have also established the microsurgery technique to ablate the sclerotic hippocampal tissue in epileptic rodents along the lines of hippocampal ablation in TLE patients. To attain minimal invasiveness, we have injected a cytotoxic agent in the hippocampus through a microcannula. This technique enables access to this deep-brain region in a closed-skull approach, without manipulating the surrounding brain areas.
We have also begun developing the computational models and signal processing tools that will aid in designing machine learning and AI algorithms. Based on supervised epileptiform events labelling, we have designed unsupervised event detection algorithms as well as Neural Mass Models (NMMs) for AI-based prediction of dynamic changes in brain network states. The developed NMMs have also been included in the computer simulation framework to simulate the dynamics of graft, host and NCS, alone and in combination.
The neuromorphic concept developed in HERMES is the first two-way learning system, where the biological and artificial synapses learn from each other. The AFE design brings further novelty to the NCS and broadly to neural recording device engineering by enabling detection of action potentials and local field potentials in a wide amplitude range with >250 times input impedance improvement versus other structures.
Aiming at the in vivo setting, to overcome the invasiveness of hippocampal ablation, we have established an effective closed-skull neurosurgery strategy which we foresee will outclass current organoid grafting procedures limited to the cortical surface. The approach may also be highly relevant for the envisioned combined ablation/replacement of damaged brain tissue in humans. For NCS implant, the modelled microprobe is based on the previously unconsidered combination of biocompatible substrate and encapsulation polymers. Simulations show higher flexibility than commercial silicon counterparts and greater aspect ratio than the state-of-the-art, heralding the feasibility of truly long-term biomedical brain implants.
The methods for detection/prediction of brain-state changes and the modular simulation framework to realistically model multiple components interplay bring additional novelty in being based on a general framework rather than on detailed brain function biophysics, for which we expect them to find wider use in neuroscience and bio-ICT convergence research as well as in the clinical setting for diagnosis tools and brain prosthetics design.
By the end of the project, we expect to establish innovative bio/neurotechnology that will impact on biomaterials, brain tissue bioengineering, neuromorphic systems design, deep brain recording, machine learning/AI techniques, and, ultimately, enhanced brain regeneration overall. In aiming at shifting the mindset of biomedical interventions from treating to healing, enhanced regenerative medicine represents a giant conceptual leap for healthcare and society. For this, we will push towards highly scalable, versatile intelligent biohybrids of non-disease-specific design to set the stage for their future implementation beyond epilepsy. The disruptive nature of such advancement will potentially generate major returns on health care and society by bringing previously unimaginable possibilities to cure disorders that represent today a global major burden of disease.