Mouthcheck, uma ferramenta de apoio no diagnóstico de câncer de boca utilizando redes neurais convolucionais

Abstract

This study aimed to develop a visual tool implementing a Convolutional Neural Network (CNN) model to assist in the early diagnosis of oral cancer. A total of 950 oral mucosa images were used, obtained from the "ORAL CANCER DATASET,"which includes labeled images indicating the presence or absence of cancer. Three CNN models, based on the VGG16, InceptionV3, and ResNet50 architectures, were trained using transfer learning techniques to reduce training time. The results showed that the ResNet50 model demonstrated superior performance in accuracy, precision, recall, and F1-Score metrics. Future work may explore expanding the dataset and implementing new features in the interface. This study contributes to the healthcare field by offering a simple solution that can assist in early diagnosis, potentially improving clinical outcomes.