Validating the laminar population analysis for decomposing cortical multielectrode data on a model of the mouse primary visual cortex
Year: 2019
Type: dissertation
Abstract: Recent technological innovation has led to a vast increase in the amount neural data that can be collected at once, and spawned the need for new methods to analyze these large, multidimensional data sets. Laminar multielectrodes measures the extracellular potential (ECP) and allows recording of the activity from thousands of neurons simultaneously across the cortical layers. However, since many sources contribute to the measured ECP it can also be hard to interpret. Traditionally, the high-frequency (above 500 Hz) part of the ECP, referred to as multi-unit activity (MUA), has been used to extract spikes, while the low-frequency (below 500 Hz) part, referred to as the local field potential (LFP), has largely been ignored. Efforts to improve interpretability of the LFP by decomposing it into the different sources have often been hampered by the lack of suitable tools for performing this decomposition. One method developed to remedy this is the laminar population analysis (LPA), where the MUA and LFP are utilized jointly in the decomposition. Here, the adequacy of this method was assessed by applying it on a model of the primary visual cortex currently under development at the Allen Institute for Brain Science, and its performance was evaluated by comparing it to the performance of the standard decomposition method principal components analysis (PCA). With LPA, three out of the five laminar populations were correctly identified and separated from the other layers, while no single layer was identified and separated with PCA. Neither LPA nor PCA convincingly separated the contributions of the different laminar populations to the LFP. All the populations that LPA partitioned the LFP into were dominated by excitatory layer 5 and layer 6 cells. The only indication of a possible separation of the different contributions was that one population was more influenced by layer 5 cells than cells in other layers. With PCA, all major sources to the LFP were allocated to the first component. The inefficacy of both approaches in the decomposition of the LFP could be due to artefacts of this model, such as overly synchronous firing across layers, and the dominance of the LFP by contributions from layer 5 and layer 6 excitatory cells.
Source: NVA (Nasjonalt vitenarkiv)
Author: Atle Eskeland Rimehaug
Institution: Norwegian University of Science and Technology
Language: English
Subfield: Cognitive Neuroscience
Field: Neuroscience
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W7062576722
