Genov, R. | bioRxiv Science

bioRxiv2022

OpenMEA: Open-Source Microelectrode Array Platform for Bioelectronic Interfacing

Bioelectronic interfaces have the potential to revolutionize the treatment of medical disorders and augment physiology. Implantable devices such as pacemakers and deep brain stimulators have already been deployed to control activity in diseases including Parkinsons disease and epilepsy. These devices typically operate by delivering electrical stimulation at pre-programmed intervals (known as open-loop stimulation). Recent advances in machine learning and low-power integrated circuits have led to the emergence of personalized medical devices that monitor the users state and stimulate in response to measured biological activity (known as closed-loop stimulation). There are two key questions that require fundamental research to achieve breakthroughs in personalized devices: 1) What biomarkers and algorithms are best suited to detecting biological states (e.g. seizures in epilepsy)? and 2) What types of electrical stimuli are optimal for controlling these states? The answer to these questions can be explored in vitro using multielectrode array (MEA) systems that interface with biological tissue with reduced experimental complexity, better reproducibility, and fewer confounding variables present in whole organisms. However, existing MEA systems have functional limitations and closed-source designs that prevent researchers from developing improvements. This paper introduces OpenMEA, an open-source platform for closed-loop bioelectronics research. OpenMEA includes designs for the components necessary to build a benchtop in vitro laboratory, including electrophysiological recording and stimulation electronics, a microfluidic perfusion system, and physical designs for multielectrode arrays. The system is demonstrated with the electrical recording and stimulation of epileptogenic human and rodent brain slices. The aim of OpenMEA is to democratize bioelectronic research tools to accelerate the deployment of devices for the treatment of disorders and beyond.

bioengineering↗

bioRxiv2022

Resource-efficient Neural Network Architectures forClassifying Nerve Cuff Recordings on Implantable Devices

BackgroundClosed-loop control of functional electrical stimulation involves using recorded nerve signals to make decisions regarding nerve stimulation in real-time. Surgically implanted devices that can implement this strategy have significant potential to restore natural movement after paralysis. Previous work demonstrated the use of convolutional neural networks (CNNs) to discriminate between activity from different neural pathways recorded by a high-density multi-contact nerve cuff electrode. Despite state-of-the-art performance, that approach required too much data storage, power and computation time for a practical implementation on surgically implanted hardware. ObjectiveTo reduce resource utilization for an implantable implementation, with a minimal performance loss for CNNs that can discriminate between neural pathways in multi-contact nerve cuff electrode recordings. MethodsNeural network (NN) architectures were evaluated on a dataset of rat sciatic nerve recordings previously collected using 56-channel (7 x 8) spiral nerve cuff electrodes to capture spatiotemporal neural activity patterns. The NNs were trained to classify individual, natural compound action potentials (nCAPs) elicited by sensory stimuli. Three architecture types were explored: the previously reported ESCAPE-NET, a fully convolutional network, and a recurrent neural network. Variations of each architecture yielded NNs with a range in the number of weights and required floating-point operations (FLOPs). Each NN was evaluated based on F1-score and resource requirements. ResultsNNs were identified that, when compared to ESCAPE-NET, required 1,132-1,787x fewer weights, 389-995x less memory, and 6-11,073x fewer FLOPs, while maintaining macro F1-scores of 0.70-0.71 compared to a baseline of 0.75. Memory requirements range from 22.69 KB to 58.11 KB, falling within the range of on-chip memory sizes from several published deep learning accelerators fabricated in 65nm ASIC technology. ConclusionReduced versions of ESCAPE-NET require significantly fewer resources without significant accuracy loss, thus can be more easily incorporated into a surgically implantable device that performs closed-loop real-time responsive neural stimulation.

neuroscience↗