Methods to Enhance Information Extraction from Microelectrode Array Measurements of Neuronal Networks

For the last couple of decades, hand-in-hand progresses in stem-cell technologies and culturing neuronal cells together with advances in microelectrode array (MEA) technology have enabled more efficient biological models. Thus, understanding the neuronal behavior in general or realizing some particular disease model by studying neuronal responses to pharmacological or neurotoxical assays has become more achievable.Moreover, with the widespread practical usage of MEA technology, a vast amount of new types of data has been collected to be analyzed. Conventionally, MEA data from neuronal networks have been analyzed, e.g., with the methods using predefined parameters and suitable for analyzing only specific neuronal behaviors or by considering only a portion of the data such as extracted extracellular action potentials (EAPs). Therefore, in addition to the current analysis methods, novel methods and newly acquired measures are needed to understand the new models. In fact, we hypothesized that existing measurement data carry a lot more information than is considered at present. In this thesis, we proposed novel methods and measures to increase the information which we can extract from MEA recordings; thus, we hope these to contribute to better understanding of neuronal behaviors and interactions.Firstly, to analyze firing properties of neuronal ensembles, we developed a method which identifies bursts based on spiking behavior of recordings; thus, the method is feasible for the cultures with variable firing dynamics. The developed method was also designed to process a large amount of data automatically for statistical justification. Therefore, we increased the analysis power in the subsequent analyses in comparison to the existing burst detection methods which are using pre-defined and strict definitions.Subsequently, we proposed novel metrics to evaluate and quantify the information content of the bursts. Entropy-based measures were employed for quantifying bursts according to their self-similarity and spectral uniformity. We showed that different types of bursts can be distinguished using entropy-based measures. Also, the joint analysis of bursts and action potential waveforms were proposed to obtain a novel type of information, i.e., spike type compositions of bursts. We presented that the spike type compositions of bursts would change under different pharmacological applications.In addition, we developed a novel method to calculate synchronization between neuronal ensembles by evaluating their time variant spectral distributions: For that, we assessed correlations of the spectral entropy (CorSE). We showed that CorSE was able to estimate synchronicity by studying both local field potentials (LFPs) and extracellular action potentials (EAPs); thus, we could contribute to understanding synchronicity between neuronal ensembles which also don’t exhibit detectable EAPs.In conclusion, motivated by the recent popularity of MEA usage in the neuroscience field, we developed novel and enhanced methods to derive new types of information. We showed that by using our developed methods one could extract additional information from MEA recordings. As a result, the proposed methods and metrics would enhance the analysis efficiency of the microelectrode array measurement based studies and provide different viewpoints for the analyses. The derived novel information would contribute to interpreting neuronal signals recorded from a single or multiple recording locations. Consequently, methods presented in this thesis are important complements to the existing methods to understand neuronal behavior and population-wise neuronal interactions