| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 5.21 MB | Adobe PDF |
Authors
Abstract(s)
A deteção do cancro de mama por micro-ondas é uma técnica bastante recente,
sendo baseada na diferença das propriedades dielétricas entre os tecidos saudáveis
e os tecidos malignos. Embora esta técnica apresente bastantes vantagens
em relação a outras técnicas convencionais, ainda apresenta algumas limitações.
Assim, esta dissertação tem como objetivo contribuir para o desenvolvimento da
técnica de imagem médica por micro-ondas. Foram implementados quatro algoritmos
diferentes de processamento de sinal que permitem remover os artefactos
indesejados nos sinais de um sistema de imagem médica baseado em microondas.
O artefacto mais conhecido e o mais crítico de todos é a reflexão ocorrida
na superfície da pele, que pode ser de várias ordens de magnitude superior à reflexão
ocorrida no tumor. Para além disso, são implementados dois algoritmos
de formação de imagem médica baseados em métodos de beamforming.
Por fim, para testar o correto funcionamento de cada um dos algoritmos e analisar
o seu desempenho, de forma a que possa ser feita uma comparação entre cada um
dos algoritmos, é utilizada uma variedade de phantoms numéricos, baseados em
modelos FDTD (Finite Difference Time Domain) anatomicamente realistas.
Os resultados permitiram concluir que para alguns algoritmos o artefacto é removido
com sucesso, sendo que, em qualquer um dos cenários simulados foi
possível detetar o tumor com sucesso. No entanto, verificou-se a existência de
um desvio de alguns milímetros da sua localização real, o que indica que ainda
existem alguns desafios a superar no âmbito da formação da imagem médica por
micro-ondas.
The detection of breast cancer by microwave imaging is a fairly recent technique, based on the difference in dielectric properties between healthy tissues and malignant tissues. Although this technique has many advantages over other conventional techniques, it still has some limitations. Thus, this dissertation aims to contribute to the development of microwave imaging technique. Four different signal processing algorithms have been implemented that allow the removal of unwanted artifacts from the signals of a microwave-based medical imaging system. The best-known and most critical artifact is the reflection on the surface of the skin, which can be of several orders of magnitude higher than the reflection occurred in the tumor. In addition, two medical imaging algorithms are implemented based on beamforming techniques. Finally, to test the correct functioning of each of the algorithms and to analyze their performance, so that a comparison can be made between each of the algorithms implemented, a variety of numerical breast phantoms are used in anatomically realistic FDTD (Finite Difference Time Domain) models. The results allowed to conclude that for some algorithms the artifact is successfully removed, and in any of the simulated scenarios it was possible to detect the tumor successfully. However, there has been a deviation of some millimeters from its actual location, which indicates that there are still some challenges to overcome in microwave imaging techniques.
The detection of breast cancer by microwave imaging is a fairly recent technique, based on the difference in dielectric properties between healthy tissues and malignant tissues. Although this technique has many advantages over other conventional techniques, it still has some limitations. Thus, this dissertation aims to contribute to the development of microwave imaging technique. Four different signal processing algorithms have been implemented that allow the removal of unwanted artifacts from the signals of a microwave-based medical imaging system. The best-known and most critical artifact is the reflection on the surface of the skin, which can be of several orders of magnitude higher than the reflection occurred in the tumor. In addition, two medical imaging algorithms are implemented based on beamforming techniques. Finally, to test the correct functioning of each of the algorithms and to analyze their performance, so that a comparison can be made between each of the algorithms implemented, a variety of numerical breast phantoms are used in anatomically realistic FDTD (Finite Difference Time Domain) models. The results allowed to conclude that for some algorithms the artifact is successfully removed, and in any of the simulated scenarios it was possible to detect the tumor successfully. However, there has been a deviation of some millimeters from its actual location, which indicates that there are still some challenges to overcome in microwave imaging techniques.
Description
Dissertação para obtenção do grau de mestre em Engenharia Electrónica e Telecomunicações
Keywords
Cancro da mama Breast cancer Micro-ondas Microwaves Remoção de artefactos Artifacts removal FDTD
Citation
LOPES, Pedro Manuel Osório - Deteção do cancro da mama por MWI. Lisboa: Instituto Superior de Engenharia de Lisboa, 2018. Dissertação de mestrado.
Publisher
Instituto Superior de Engenharia de Lisboa