NEURONAL CLASSIFICATION BASED ON HIGH SPATIAL AND TEMPORAL RESOLUTION EXTRACELLULAR ELECTROPHYSIOLOGICAL RECORDINGS PERFORMED USING HD-MEAS

Strategies to navigate the complexity of the brain are important for a bottom-up understanding of the function (and dysfunction) of neural circuits. The first step towards reducing complexity is to create a parts list of the individual elements comprising neural circuits. Identifying functionally distinct types of neurons enables the systematic analysis of their individual contributions to circuit function. Yet, reliable and high-throughput neuron type classification remains a challenge. Modern extracellular electrophysiological devices offer access to the activity of neural ensembles at high spatiotemporal resolution. In this study we asked if multi-scale features harvested from high-resolution extracellular electrophysiology enable reliable and high-throughput profiling of neurons into two broad functional classes: excitatory and inhibitory. We addressed this question using generic in vitro networks of rat primary dissociated hippocampal neurons grown on high-density microelectrode arrays. Using ground truth labels–assigned based on spike train correlations or molecular features–we assessed the feasibility of such a task. We also used the data set to train and test linear and nonlinear classifier models and evaluated the influence of individual features on overall performance, in order to infer what electrophysiological properties might better represent the two classes.