Investigation of in vitro neuronal activity processing using a CMOS-integrated ZrO2-based memristive crossbar

TL;DR

This study investigates how epileptiform neuronal activity affects a CMOS-integrated ZrO2 memristive crossbar's conductivity and synaptic plasticity, with implications for real-time epilepsy treatment and neurohybrid device development.

Contribution

It demonstrates the impact of epileptiform activity on memristive synaptic plasticity and noise effects, advancing neuromorphic hardware for medical and robotic applications.

Findings

Epileptiform activity reduces synaptic weight change in memristive crossbars.

Noise in neuronal signals decreases memristive response magnitude.

Results support development of real-time neurohybrid devices for epilepsy management.

Abstract

The influence of the epileptiform neuronal activity on the response of a CMOS-integrated ZrO2-based memristive crossbar and its conductivity was studied. Epileptiform neuronal activity was obtained in vitro in the hippocampus slices of laboratory mice using 4-aminopyridine experimental model. Synaptic plasticity of the memristive crossbar induced by epileptiform neuronal activity pulses was detected. Qualitatively, the results obtained in the case of normal (without pathologies) and epileptiform neuronal activity with and without noise coincide. For quantitative analysis, the value of the relative change in synaptic weight has been calculated for such important biological mechanisms of synapses as paired-pulse facilitation/depression, post-tetanic potentiation/depression, and long-term potentiation/depression. It has been shown that average value of the relative change in synaptic weight and its are smaller mainly in the case of epileptiform neuronal activity pulses. An effect of the influence of noise included in the neuronal activity was found, which consists in the fact that the current response of the memristive crossbar is smaller in the presence of noise. The results of this study can be used in the development of new generation hardware-implemented computing devices with high performance and energy efficiency for the tasks of restorative medicine and robotics. In particular, using these results, neurohybrid devices can be developed for processing epileptiform activity in real time and for its suppression.