Melanoma classification with neural networks using an unbalanced dataset of skin lesion images
The applications of Artificial Intelligence (AI) in various fields are extensive and have the potential to revolutionize various aspects of modern healthcare, for example, demonstrating promising advances in improving the accuracy and efficiency of skin cancer detection and classification. This a...
| Autor principal: | KLAUTAU, Sofia Pinheiro |
|---|---|
| Grau: | Trabalho de Curso - Graduação - Monografia |
| Publicado em: |
2025
|
| Assuntos: | |
| Acesso em linha: |
https://bdm.ufpa.br/jspui/handle/prefix/7619 |
| id |
oai:https:--bdm.ufpa.br:8443:prefix-7619 |
|---|---|
| recordtype |
dspace |
| spelling |
oai:https:--bdm.ufpa.br:8443:prefix-76192025-01-16T03:02:34Z Melanoma classification with neural networks using an unbalanced dataset of skin lesion images KLAUTAU, Sofia Pinheiro RAMALHO, Leonardo Lira http://lattes.cnpq.br/7565458988876048 MÜLLER, Ana Carolina Quintão Siravenha http://lattes.cnpq.br/4383482501456728 https://orcid.org/0000-0003-3165-1941 https://orcid.org/0000-0001-6664-9847 Machine Learning, , , , , . Dataset desbalanceado Câncer de pele Melanoma Otimização Classificação de imagens Skin cancer lesion Unbalanced dataset Melanoma Optimization Image classification CNPQ::ENGENHARIAS::ENGENHARIA BIOMEDICA The applications of Artificial Intelligence (AI) in various fields are extensive and have the potential to revolutionize various aspects of modern healthcare, for example, demonstrating promising advances in improving the accuracy and efficiency of skin cancer detection and classification. This area of study is of significant importance as it seeks to improve early identification and diagnosis of skin cancer, positively impacting patient outcomes and treatment strategies. This work describes a study carried out on the use of unbalanced datasets for the classification of images of skin cancer lesions using Artificial Neural Networks, more specifically, a dataset that has over 98% of samples belonging to the negative class. Three strategies were applied to try to mitigate the difficulties caused by the large difference in the number of images in each class, in this case, lesions that are melanoma and lesions that are not melanoma: reducing the number of samples in the dataset to balance it, applying data augmentation and applying class weights. In addition, methods for optimizing the training process of a Convolutional Neural Network are successfully applied to automate the hyperparameters selection process and the training time of models that use large neural networks as feature extractors is reduced because of it. The data augmentation and class weights adopted in this work helped the training procedure but were not enough to produce a large improvement in performance, but the latter method was applied in the best result obtained. As aplicações de Inteligência Artificial (IA) em vários campos são extensas e têm o potencial de revolucionar vários aspectos da saúde moderna, por exemplo, demonstrando avanços promissores na melhoria da precisão e eficiência da detecção e classificação do câncer de pele. Esta área de estudo é de grande importância, pois busca melhorar a identificação e o diagnóstico precoce do câncer de pele, impactando positivamente os resultados dos pacientes e as estratégias de tratamento. Este trabalho descreve um estudo realizado sobre o uso de conjuntos de dados desbalanceados para a classificação de imagens de lesões de câncer de pele usando Redes Neurais Artificiais, mais especificamente, um conjunto de dados que possui mais de 98% de amostras pertencentes à classe negativa. Três estratégias foram aplicadas para tentar mitigar as dificuldades causadas pela grande diferença no número de imagens em cada classe, no caso, lesões que são melanoma e lesões que não são melanoma: reduzir o número de amostras no conjunto de dados para equilibrá-lo , aplicando aumento de dados e aplicando pesos de classe. Além disso, métodos para otimizar o processo de treinamento de uma Rede Neural Convolucional são aplicados com sucesso para automatizar o processo de seleção de hiperparâmetros e o tempo de treinamento de modelos que usam grandes redes neurais como extratores de características é reduzido por causa disso. O aumento de dados e pesos de classe adotados neste trabalho ajudaram no procedimento de treinamento, mas não foram suficientes para produzir uma grande melhoria no desempenho, porém o último método foi aplicado no melhor resultado obtido 2025-01-15T15:48:36Z 2025-01-15T15:48:36Z 2023-07-17 Trabalho de Curso - Graduação - Monografia KLAUTAU, Sofia Pinheiro. Melanoma classification with neural networks using an unbalanced dataset of skin lesion images. Orientador: Leonardo Lira Ramalho. 2023. 63 f. Trabalho de Curso (Bacharelado em Engenharia Elétrica e Biomédica) – Faculdade de Engenharia Elétrica e Biomédica, Instituto de Tecnologia, Universidade Federal do Pará, Belém, 2023. Disponível em: https://bdm.ufpa.br/jspui/handle/prefix/7619. Acesso em:. https://bdm.ufpa.br/jspui/handle/prefix/7619 Acesso Aberto Disponível na internet via correio eletrônico: bibliotecaitec@ufpa.br |
| institution |
Biblioteca Digital de Monografias - UFPA |
| collection |
MonografiaUFPA |
| topic |
Machine Learning, , , , , . Dataset desbalanceado Câncer de pele Melanoma Otimização Classificação de imagens Skin cancer lesion Unbalanced dataset Melanoma Optimization Image classification CNPQ::ENGENHARIAS::ENGENHARIA BIOMEDICA |
| spellingShingle |
Machine Learning, , , , , . Dataset desbalanceado Câncer de pele Melanoma Otimização Classificação de imagens Skin cancer lesion Unbalanced dataset Melanoma Optimization Image classification CNPQ::ENGENHARIAS::ENGENHARIA BIOMEDICA KLAUTAU, Sofia Pinheiro Melanoma classification with neural networks using an unbalanced dataset of skin lesion images |
| topic_facet |
Machine Learning, , , , , . Dataset desbalanceado Câncer de pele Melanoma Otimização Classificação de imagens Skin cancer lesion Unbalanced dataset Melanoma Optimization Image classification CNPQ::ENGENHARIAS::ENGENHARIA BIOMEDICA |
| description |
The applications of Artificial Intelligence (AI) in various fields are extensive and have the
potential to revolutionize various aspects of modern healthcare, for example, demonstrating
promising advances in improving the accuracy and efficiency of skin cancer detection and
classification. This area of study is of significant importance as it seeks to improve early
identification and diagnosis of skin cancer, positively impacting patient outcomes and treatment
strategies. This work describes a study carried out on the use of unbalanced datasets for the
classification of images of skin cancer lesions using Artificial Neural Networks, more specifically,
a dataset that has over 98% of samples belonging to the negative class. Three strategies were
applied to try to mitigate the difficulties caused by the large difference in the number of images
in each class, in this case, lesions that are melanoma and lesions that are not melanoma: reducing
the number of samples in the dataset to balance it, applying data augmentation and applying
class weights. In addition, methods for optimizing the training process of a Convolutional Neural
Network are successfully applied to automate the hyperparameters selection process and the
training time of models that use large neural networks as feature extractors is reduced because of
it. The data augmentation and class weights adopted in this work helped the training procedure
but were not enough to produce a large improvement in performance, but the latter method was
applied in the best result obtained. |
| author_additional |
RAMALHO, Leonardo Lira |
| author_additionalStr |
RAMALHO, Leonardo Lira |
| format |
Trabalho de Curso - Graduação - Monografia |
| author |
KLAUTAU, Sofia Pinheiro |
| title |
Melanoma classification with neural networks using an unbalanced dataset of skin lesion images |
| title_short |
Melanoma classification with neural networks using an unbalanced dataset of skin lesion images |
| title_full |
Melanoma classification with neural networks using an unbalanced dataset of skin lesion images |
| title_fullStr |
Melanoma classification with neural networks using an unbalanced dataset of skin lesion images |
| title_full_unstemmed |
Melanoma classification with neural networks using an unbalanced dataset of skin lesion images |
| title_sort |
melanoma classification with neural networks using an unbalanced dataset of skin lesion images |
| publishDate |
2025 |
| url |
https://bdm.ufpa.br/jspui/handle/prefix/7619 |
| _version_ |
1829093140918697984 |
| score |
11.753896 |