| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 10.36 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
Esta tese explora a aplicação de redes neurais convolucionais (CNNs) para a classificação de
úlceras do pé diabético, utilizando as arquiteturas VGG16, VGG19 e MobileNetV2. O objetivo
principal é desenvolver e comparar modelos de deep learning capazes de identificar com
precisão áreas lesionadas em imagens clínicas de pés diabéticos, ajudando na prevenção e no
tratamento eficaz das úlceras. Foi realizado um estudo com base num conjunto de dados de
imagens anotadas, avaliando o desempenho dos modelos em termos da accuracy, precisão,
recall e F1-Score. A VGG19 destacou-se com uma accuracy de 93%, evidenciando uma
capacidade superior de localizar lesões em imagens complexas, com ativações mais focadas nas
áreas relevantes. A MobileNetV2, por outro lado, apresentou um bom desempenho em termos
de eficiência computacional, sendo adequada para dispositivos móveis e ambientes com
restrições de hardware, embora tenha apresentado um valor de accuracy ligeiramente inferior
em relação às VGGs. O estudo também discute as limitações de cada arquitetura, como a maior
tendência ao overfitting nos modelos mais profundos e a menor capacidade de abstração de
detalhes clínicos no MobileNetV2. Os resultados indicam que o uso de CNNs tem grande
potencial no diagnóstico clínico assistido por imagem, especialmente para patologias como o
pé diabético, onde a deteção precoce e precisa é crucial para evitar amputações.
This thesis investigates the application of convolutional neural networks (CNNs) for the classification of diabetic foot ulcers, using the VGG16, VGG19, and MobileNetV2 architectures. The primary objective is to develop and compare deep learning models capable of accurately identifying ulcerated regions in clinical images of diabetic feet, thereby aiding in the prevention and effective treatment of foot ulcers. A comprehensive study was conducted using an annotated dataset of medical images, evaluating the performance of the models in terms of accuracy, precision, recall, and F1-score. VGG19 achieved the highest accuracy at 93%, demonstrating superior ability to focus activations on relevant lesion areas in complex images. MobileNetV2, while slightly less accurate, excelled in computational efficiency, making it a suitable choice for mobile devices and environments with hardware constraints. The study also highlights the limitations of each architecture, such as increased risk of overfitting in deeper models and the lower capability of MobileNetV2 to capture fine clinical details. These findings suggest that CNNs hold significant potential in computer-aided clinical diagnosis, particularly in the early and precise detection of diabetic foot ulcers, where timely intervention is crucial to prevent amputations.
This thesis investigates the application of convolutional neural networks (CNNs) for the classification of diabetic foot ulcers, using the VGG16, VGG19, and MobileNetV2 architectures. The primary objective is to develop and compare deep learning models capable of accurately identifying ulcerated regions in clinical images of diabetic feet, thereby aiding in the prevention and effective treatment of foot ulcers. A comprehensive study was conducted using an annotated dataset of medical images, evaluating the performance of the models in terms of accuracy, precision, recall, and F1-score. VGG19 achieved the highest accuracy at 93%, demonstrating superior ability to focus activations on relevant lesion areas in complex images. MobileNetV2, while slightly less accurate, excelled in computational efficiency, making it a suitable choice for mobile devices and environments with hardware constraints. The study also highlights the limitations of each architecture, such as increased risk of overfitting in deeper models and the lower capability of MobileNetV2 to capture fine clinical details. These findings suggest that CNNs hold significant potential in computer-aided clinical diagnosis, particularly in the early and precise detection of diabetic foot ulcers, where timely intervention is crucial to prevent amputations.
Description
Keywords
Convolutional neural networks Medical image classification Diabetic foot ulcers Deep learning Redes neurais convolucionais Úlceras do pé diabético Classificação de imagens médicas