Multimodal Microelectrode Failure Analysis Reveals Complex Relationship at the Neural Interface @ ECS

Post date: Oct 2, 2017 9:58:57 AM

Penetrating microelectrode arrays that can record extracellular action potentials from small, targeted groups of neurons are critical for basic neuroscience research and emerging clinical applications. However, these electrode devices suffer from reliability and variability issues which impact their performance on the order of months to years. The failure mechanisms of these electrodes are understood to be a complex combination of the biotic and abiotic failure modes.

The breaching of the blood–brain barrier (BBB) to insert devices triggers a cascade of biochemical pathways resulting in complex molecular and cellular responses to implanted devices. Molecular and cellular changes in the microenvironment surrounding an implant include the introduction of mechanical strain, BBB leakage, activation of glial cells, loss of perfusion, secondary metabolic injury, and neuronal degeneration. The resulting inflammation is a key hypothesized cause of neural recording failure. However, previous attempt so directly correlate recording performance, to impedance, and to histological outcomes have led counter-intuitive and sometimes conflicting outcomes.

One reason is that many neurons remain quiescent during anesthetized or resting-state conditions. We previously demonstrated this by visually evoked stimulation paradigms of the contralateral eye in order to evaluate chronic recording performance of linear silicon electrode in the primary visual cortex. Additional, multiphoton analysis using GCaMP6 transgenic animals further confirmed these results. More recently, there has been a growing interesting recording during awake free-roaming conditions in the primary motor cortex in order to avoid resting-state related quiescent activity. However, this in turn leads to increases in Lenz’s Law related artefacts that have the same time constants and waveform shapes as action potentials in rodents, but not NHP. While behaviorally training animals to remain immobile could improve outcomes, it also introduces the potential for Experimenter Expectancy Effect bias on the outcomes.

The visual stimulation paradigm enable the use of current source density analysis to electrophysiologically identify Layer II/III, IV, and V in the cortex. This, in turn, allowed correlation of electrophysiological layers to the histological layers based on section depth and the differences in neural morphology and density. Our findings from electrophysiology, impedance spectroscopy, and post-mortem histology demonstrate a very poor relationship between histology and impedance to electrophysiology. For example, tissue with low-levels of glial encapsulation, healthy neuronal proximity, and low impedance can still have poor recording performance, even with neural activity is behaviorally driven.

Even when histology confirms a perfect tissue interface, cracking or delamination of insulation on the microelectrode has been linked to a drop in impedance and a loss of recording failure. In contrast, cracking of the electrical trace and delamination of the recording site has been linked to recording failure through a jump in electrical impedance. As such, several modes of mechanical failure of chronically implanted planar silicon electrodes were found that result in degradation and/or loss of recording. Our findings highlight the importance of strains and material properties of various subcomponents within an electrode array and the poor reliability of determining electrode viability through electrochemical impedance spectroscopy.

Interestingly, we discovered in a number of situations that even with good neural density, uncompromised electrode material, and good impedances, recording performance can sometimes completely degrade. New multimodal analysis demonstrates the importance of capturing dynamic information, such as with in vivo multiphoton study, and that the presence of neurons does not guarantee functional neural activity over time. We further demonstrate that the foundation of assumptions and simplification made in the field for neural interface research are not true or incomplete. To solve the longstanding chronic neural interface problem, we need to first understand the complexity of the problem.

    • © 2017 ECS - The Electrochemical Society