Quantificação e agrupamento de neurônios in vitro usando mapas auto-organizáveis em matriz de microeletrodos

• Ano de defesa: 2020
• Autor(a) principal: Campos, José Roberto
• Orientador(a): Saito, José Hiroki
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de São Carlos; Câmpus São Carlos
• Programa de Pós-Graduação: Programa de Pós-Graduação em Ciência da Computação - PPGCC
• Palavras-chave em Português: Matriz de microeletrodos; Neurocomputação; Classificação
• Palavras-chave em Inglês: Multielectrode array; Neurocomputation; Classification; Spikes; Bursts
• Área do conhecimento CNPq: CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO
• Link de acesso: https://repositorio.ufscar.br/handle/20.500.14289/12794

Understanding how information is encoded in large neuronal networks is one of the biggest challenges in neuroscience. A possible approach to investigate the information processing capabilities of neuronal nets is given by the use of dissociated neuronal cultures attached to the Microelectrode Array. For instance, cortical neurons dissociated from rat embryos cultivated in vitro in MEA show characteristic patterns of electrophysiological activities, ranging from isolated peaks in the early days up to highly synchronized bursts after 3-4 weeks. In this context it is important to know the spatial distribution function of neurons and, for that the microscopic images are an alternative used to find out the positioning of the neuron cells on the glass substrate of the MEA. However, it is worth noting that microscopy is not always performed mainly at short intervals during signal recording. Of course there is a lack in the context of methods capable to perform such functions without impairing the measurement of spontaneous signals produced by the neuronal networks. To minimize this problem, it is proposed in this thesis a framework for quantifying and classifying neuron cells and thus estimate the topological arrangement of the culture, by identifying spikes and bursts, on the MEA electrodes. The framework developed for mapping the topology of neurons, using Self-Organizing Map (SOM) neural network, is applied to the recording signals of the experiment entitled Culture 371 of dissociated neuronal cells, carried out at the Laboratory of Neuroengineering and Bionanotechnology at the University of Genoa, Italy. To validate the procedure, a topological map obtained as result of this thesis is compared with a microscopic image obtained on day 38 of maturation of Culture 371. Considering the majority number of MEA electrodes, the Chi-Square test resulted in confront of this validation. In addition, it is applied in the topological maps the classification the neurons observed around the microelectrodes using the analysis and separation of the registered spike waveforms.