Identificação de tumores cerebrais por meio do modelo de contornos ativos e algoritmos genéticos
Resumo
Resumo: Embora a identificação de tumores cerebrais seja uma tarefa difícil no diagnóstico
médico, avanços significativos têm sido verificados devido à evolução das técnicas de aquisição de imagens médicas. Por outro lado, a computação gráfica oferece técnicas que possibilitam a adequada manipulação destas imagens. Entre os diversos métodos existentes, o modelo de contornos ativos, também conhecido como snakes, vem sendo amplamente difundido no processo de segmentação e extração de estruturas de interesse no contexto médico. Este trabalho investiga o método de snakes propondo sua utilização em conjunto com algoritmos genéticos para resolver as restrições descritas na literatura. Os resultados obtidos demonstram o potencial desta abordagem e incentivam a continuação de trabalhos de pesquisa no sentido de ampliar sua contribuição no diagnóstico médico avançado. Abstract:The identification of brain tumours is a complex task in the context of diagnosis in medicine. However, significant advances can be observed in this field due to the progress of medical image acquisition techniques. On the other hand, computer graphics offers techniques which provide excellent tools for manipulating these images. In particular, the active contour model, also known as "snakes", has been playing an important role in the process of segmentation and extraction of structures of interest in clinical practice. This work investigates the "snakes" method and proposes its use in conjunction with genetic algorithms to address the limitations reported in the literature. The results thus obtained demonstrate the potential of this technique and encourage further research work in order to increase its contribution in advanced medical diagnosis.
Collections
- Teses & Dissertações [10146]