| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 1.9 MB | Adobe PDF |
Abstract(s)
O Acidente Vascular Cerebral (AVC) tem um grande impacto na sociedade moderna, sendo a principal causa de morte em Portugal. A aterosclerose é uma doença crónica progressiva do sistema vascular, com alterações celulares e metabólicas na parede arterial. Na doença aterosclerótica podem ser desenvolvidas placas ateroscleróticas vulneráveis, aumentado assim, o risco de eventos cardiovasculares, sendo possível, através de características especificas, classificar de forma qualitativa a sua vulnerabilidade. A ultrassonografia bidimensional (2D) carotídea é muito utilizada no diagnóstico da doença aterosclerótica, por ser um método de imagem não invasivo, pouco dispendioso e por fornecer imagens de alta resolução em tempo real. No entanto, na prática clínica, a deteção e a caracterização das placas ateroscleróticas são realizadas subjetivamente, sendo por isso considerada uma técnica dependente do operador. Nesta tese é desenvolvido um sistema de diagnóstico assistido por computador, em linguagem Python, totalmente automático e com precisão, na deteção dos limites anatómicos da placa aterosclerótica nas artérias carótidas, através de imagens 2D obtidas em ultrassonografia modo B. Para além disso, o algoritmo concebido para o processamento de imagem digital tem a capacidade de caracterizar a placa aterosclerótica de acordo com o grau de reflexão dos ultrassons e homogeneidade. O algoritmo desenvolvido foi aplicado em 4 imagens de casos clínicos diferentes. Em 3 casos foi possível detetar e caracterizar automaticamente as diferentes placas, sendo que no quarto caso a deteção foi parcial, evidenciando limitação na avaliação de placas mais hipoecogénicas. Os resultados demonstraram que a aplicação deste algoritmo em linguagem Python pode ser utilizado para identificar e caracterizar placas ateroscleróticas carotídeas. Aliado à avaliação clínica, este método permite aumentar a precisão e reprodutibilidade da técnica, e auxiliar na melhor decisão terapêutica e prognóstico do doente.
ABSTRACT - Stroke has a big impact on modern society, being the main cause of death in Portugal. Atherosclerosis is a progressive chronic disease of the vascular system, with cellular and metabolic changes in the arterial wall. In atherosclerotic disease, vulnerable atherosclerotic plaques can be developed, thus increasing the risk of cardiovascular events. It is possible, through specific characteristics, to qualitatively classify their vulnerability. Carotid ultrasonography is widely used in the diagnosis of atherosclerotic disease, as it is a non-invasive, inexpensive imaging method and provides high-resolution images in real-time. However, the detection and segmentation of atherosclerotic plaques is subjective and is therefore considered an operator-dependent technique. In this thesis, we develop a computer-aided, fully automatic and accurate Python-based diagnostic system for the detection of the anatomical limits of the atherosclerotic plaque in the carotid arteries through two-dimensional images obtained in B-mode ultrasonography. In addition, the algorithm designed for digital image processing has the ability to characterize the atherosclerotic plaque according to the level of reflection of the ultrasound and homogeneity. The algorithm developed was applied in 4 images of different clinical cases. In 3 cases it was possible to automatically detect and characterize the different plaques, and in the fourth case, the detection was partial, evidencing limitation in the evaluation of more hypoechogenic plaques. The results demonstrated that the application of this algorithm in Python language can be used to identify and characterize carotid atherosclerotic plaques. Together with clinical evaluation, this method allows us to increase the precision and reproducibility of the technique and to help in the better therapeutic decision and prognosis of the patient.
ABSTRACT - Stroke has a big impact on modern society, being the main cause of death in Portugal. Atherosclerosis is a progressive chronic disease of the vascular system, with cellular and metabolic changes in the arterial wall. In atherosclerotic disease, vulnerable atherosclerotic plaques can be developed, thus increasing the risk of cardiovascular events. It is possible, through specific characteristics, to qualitatively classify their vulnerability. Carotid ultrasonography is widely used in the diagnosis of atherosclerotic disease, as it is a non-invasive, inexpensive imaging method and provides high-resolution images in real-time. However, the detection and segmentation of atherosclerotic plaques is subjective and is therefore considered an operator-dependent technique. In this thesis, we develop a computer-aided, fully automatic and accurate Python-based diagnostic system for the detection of the anatomical limits of the atherosclerotic plaque in the carotid arteries through two-dimensional images obtained in B-mode ultrasonography. In addition, the algorithm designed for digital image processing has the ability to characterize the atherosclerotic plaque according to the level of reflection of the ultrasound and homogeneity. The algorithm developed was applied in 4 images of different clinical cases. In 3 cases it was possible to automatically detect and characterize the different plaques, and in the fourth case, the detection was partial, evidencing limitation in the evaluation of more hypoechogenic plaques. The results demonstrated that the application of this algorithm in Python language can be used to identify and characterize carotid atherosclerotic plaques. Together with clinical evaluation, this method allows us to increase the precision and reproducibility of the technique and to help in the better therapeutic decision and prognosis of the patient.
Description
Mestrado em Engenharia Biomédica
Keywords
AVC Acidente vascular cerebral Arteriosclerose Ultrassonografia Processamento de imagem Imagem digital Deteção automática CVA Stroke Atherosclerosis Ultrasonography Image processing Digital image Automatic detection
Citation
Cardoso MM. Deteção automática e caracterização da placa aterosclerótica através da imagem ultrassonográfica [dissertation]. Lisboa: Escola Superior de Tecnologia da Saúde de Lisboa - Instituto Superior de Engenharia de Lisboa / Instituto Politécnico de Lisboa; 2018.
Publisher
Instituto Politécnico de Lisboa, Escola Superior de Tecnologia da Saúde de Lisboa