Orientador: Aparecido Augusto de Carvalho / Resumo: O desvio lateral da coluna e a escoliose alteram o equilíbrio corporal de uma pessoa e a distribuição de seu peso nos pés. Atualmente com o auxílio do baropodômetro é possível medir a distribuição do peso corporal nos pés, trazendo inovação no que concerne sobre os impactos da escoliose nos mesmos. As alterações da coluna vertebral não são visíveis pelo baropodômetro, logo apenas mensurando a pressão dos pés não é possível determinar uma escoliose e seus possíveis ângulos. Dessa forma, adota-se o objetivo de realizar três redes neurais para classificação de escolioses com dados obtidos pelo baropodômetro do Laboratório de Instrumentação e Engenharia Biomédica (LIEB). No desenvolvimento das redes foram observadas vinte e cinco mil redes neurais feitas para cada proposta, sendo a rede neural A dividida em dois grupos que classificam o desvio lateral A1 (0º a 9º) e a escoliose A2 (10º a 20º) , a rede B foi dividida em dois grupos, B1 (10º a 13º) e B2 (14º a 20º) e a rede C1 que abrange o grupos A1, B1 e B2. A rede A (1,2) obteve uma acurácia média de 70,06%, a rede B (1,2) teve uma a acurácia média em 73,6% e a rede C (1,2,3) classificou em média 56,5% dos dados corretamente. Com os resultados obtidos conclui-se que uma classificação entre três grupos é inviável e a rede A e B podem ser utilizadas como métodos para acompanhamento de evolução ao longo do tempo. / Abstract: The lateral deviation of the spine and scoliosis alter a person's body balance and the distribution of his weight in the feet. Nowadays, with the help of the baropodometer, it is possible to measure the distribution of body weight in the feet, bringing innovation in what concerns the impact of scoliosis on them. The changes in the spine are not visible by the baropodometer, so just by measuring the pressure of the feet it is not possible to determine a scoliosis and its possible angles. Thus, we adopted the objective of performing three neural networks to classify scoliosis with data obtained by the Baropodometer of the Laboratory of Instrumentation and Biomedical Engineering (LIEB). In the development of the networks twenty-five thousand neural networks were made for each proposal, the neural network A being divided into two groups that classified the lateral deviation A1 (0º to 9º) and the scoliosis A2 (10º to 20º), the network B was divided into two groups, B1 (10º to 13º) and B2 (14º to 20º), and the C1 network encompassing groups A1, B1 and B2. The network A (1,2) obtained an average accuracy of 70,06%, the network B (1,2) had an average accuracy of 73.6% and the network C (1,2,3) classified on average 56, 5% of the data correctly. With the results obtained it is concluded that a classification between three groups is not feasible and the network A and B can be used as methods to monitor evolution over time. / Mestre

Title Evaluating the Effect of Spa Treatment for Ideopatic Scoliosis Objectives The main objective of this work is to evaluate effect of spa treatment for children with ideopatic scoliosis at age of 12-15 years. My objective was to find out the efficiency of spa treatment. I also would like to find out a long-term effect while comparing girls and boys. A long-effect means, that the spa effect will last at least until the next spa visit. The next objective was to find out how many children are affected by ideopatic scoliosis, or how many children suffer from bad body posture. I was also seeking, if there is sufficient number of spa centers for treating these kinds of diseases in Czech Republic. Methods This work has a scientific character. It is a retrospective study focused on evaluation of the effect of spa treatment for children with ideopatic scoliosis. For evaluating the rate of improvement, informations from semi-standartized tests was used. These tests are used in the spa centre in Lázně Bělohrad, Czech Republic. I extracted a selected category of patients, compared these data, and visualised them in a graphs. There are 5 physioterapists working in this centre, and all of them are using the same methods for patient monitoring. However, all data had to be standardised before working with them,...