In this arm of the research in the lab we aim to provide insights into description of human neocortical microcircuitry. We have shown before that coordinated population activity (within theta frequency range) in different cortical layers can be related to one another (be synchronized), suggesting that the oscillatory activity is a signature that modulates the activity (excitability) within the layers. We therefore are interested to investigate the functional interconnectivity between layers that allows for such a coordination between layers. To achieve that we are developing a framework for high-throughput functional mapping of single units in human cortical slices in-vitro [and in-vivo].


We hypothesize that classes of neurons in the human cortex can be identified based on: 1) Waveform features of the extracellular spikes 2) The temporal patterns of the spikes, and 3) Their laminar organization; which we call Functional or spiking phenotypes of neurons. We further hypothesize that the long-range functional connectivity between different laminae as well as intralaminar connections can be estimated using single unit activity.


Using planar multi electrode arrays, we are recording simultaneous extracellular potentials from large scale networks (i.e. covering all the cortical layers and across a few cortical columns) with high temporal resolution which allows for extracting single unit activity in a non-invasive manner compared to intracellular recordings. The experimental setup provides us with further opportunity of pharamacological intervention to influence network dynamics. In transgenic animal models we have the potential of targeted stimulation or silencing of candidate neurons (electrical or optogenetic interventions).  The extracted single unit activity in either baseline or stimulated scenarios is analyzed based on waveform features as well as temporal features. The electrodes are then registered on the histological image of the slice to find the laminar location of the identified units. Spike timing data is further analyzed to compute the relationship of units to each other based on information theoretic measures.


  1. The first critical step is to extract single unit activity from the extracellular recordings. Analysis of data generated by simultaneous recording from many electrodes is a critical step. We are developing a high throughput algorithm to produce reliable single unit activity. In particular, we are investigating the mathematical limits of reliable clustering of putative neurons from the noisy extracellular activity (that we know it has signatures of action potentials but is also formed by other sources). In the same line of research, in collaboration with Prof. Bezdek, we attempt to implement machine learning approaches such as assessing cluster tendency in the neuronal spike data in high dimensional space to the reliability of the clustering algorithm.
  2. A key element of this project is the reliable localization of the recordings (hence the identified units) on slice that registers each unit to a cortical layer. Therefore, we are developing a substitute for histology processing of the slice by using optical coherent tomography (OCT) that will let us avoid further sectioning and damages of the slice. This technique has the potential to be used on-line with recording the fresh tissue.