Proposta de segmentação de imagens histológicas de tumores da cavidade oral: Um estudo da implementação de redes neurais na análise de tumores

Abstract

Image segmentation is an important tool in the analysis, identification and diagnosis of tumors with images. Whole Slide Images (WSI) are used in the diagnosis of tumors of the oral cavity, however these images tend to be constituted of massive size and dimensions, making difficult and delaying the analysis works that are performed visually by experts. Technologies involving artificial neural networks have been integrated to perform the segmentation of regions of interest in histological images in order to help and accelerate the diagnostic processes. However, few works are carried out in this area, involving the analysis of oral cavity tumors, with some techniques only recently proposed. This work aims to study some methodologies that aim to implement the use of artificial neural networks for the segmentation of tumor regions in Whole Slide Images of the oral cavity and to reproduce some results, in the segmentation, of a new methodology proposed in the work (SANTOS, D. F.D. dos; FARIA, P. R. de; TRAVENÇOLO, B. A. N.; NASCIMENTO, M. Z. do. Automated detection of tumor regions from oral histological whole slide images using fully convolutional neural networks. Biomedical Signal Processing and Control, v. 69, 2021.) and made openly available. Some computational package compatibility problems were found due to the use of an operating system different from the one used in the development of the system, which could be circumvented and then it was obtained the segmentation of different histological images of the oral cavity, with a correct location of different tumor regions.