Segmentação Automática de Candidatos a Nódulos Pulmonares em Imagens de Tomografia Computadorizada

The Development of a Large Scale Geospatial Telecommunications Application Independent from GIS

Maria J. C. de MouraAlcilene D. de SousaIvo A. de OliveiraLaércio N. MesquitaPatrícia M. L. de L. Drumond

Eliane Z. Victorelli DiasGeovane Cayres MagalhãesGrace Kelly de C. Silva

This paper presents an algorithm for automatic segmentation of pulmonary nodules candidates in chest computed tomography images. The methodology includes acquisition images, noise elimination, segmentation of pulmonary parenchyma and segmentation pulmonary nodules candidates. The use of the filter wiener and the application of ideal threshold ensures to the algorithm a significant improvement in results, allowing to detect a greater number of nodules on the images. The tests were conducted using a set of images of the base LIDC-IDRI, containing 708 nodules. The test results showed that the algorithm located 93.08% of the nodules considered. This article presents the evolution of a very large geospatial database and application developed for telecommunications outside plant management. The needs for the construction of this system and the decisions taken during its development and consolidation phases are described. New challenges found in securing and expanding this type of system in today's market will be listed, as will be the need for a solution independent from the geographical information system (GIS) proprietary mechanisms. The alternatives examined and the approaches chosen to address the main points regarding the system evolution have resulted in a guide to similar initiatives in the area.

Caso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato:
     Mantida por: