Segmentação e volumetria automática de fígado a partir de imagens tomográficas

• Autor(a) principal: Anastácio, Rogério
• Data de Publicação: 2015
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da UFU
• Texto Completo: https://repositorio.ufu.br/handle/123456789/14099; https://doi.org/10.14393/ufu.di.2015.200
• topic: Processamento de imagens; Fígado - Tomografia; Segmentação automática do fígado; Volumetria hepática; Crescimento de região; Automatic liver segmentation; Liver volume; Region growing; CNPQ::ENGENHARIAS::ENGENHARIA BIOMEDICA

Liver cancer is highly complex to be diagnosed and treated, but is not the most frequent in Brazil, and the detection is the best way to reduce cancer mortality. And the computed tomography of the abdomen is often used in the analysis and diagnosis of hepatic lesion. The computer aided diagnosis can be used as a tool for the radiologists in the detection of liver cancer. One of the first steps of building a computer aided diagnosis for liver is the segmentation of the organ. The radiological liver contrast to the other organs in the abdominal region is low, and generally being a problem, depending on which slice is observed, to be overcome in the segmentation. The main objective of this work is to perform the automatic segmentation of the liver and calculate its volume, using the region growing algorithm segmentation. Before being performed the liver segmentation, the computed tomography slices of the abdomen are pre-processed for noise reduction and contrast enhancement. The segmentation was then performed beginning from the pre-processed slices using region growing algorithm and automatic seed launch. The seed launch area was also automated in different formats, rectangular, elliptical and circular, that influenced the performance of the segmentation technique and consequently the success rate of the hepatic volume calculation with average hit rate of 66.008% for segmentation with seed launch on rectangular area, 75.56% for the elliptical area and 94.14% for the circular area. Therefore the automatic selection of seed launching area in circular format proved to be a robust technique and with a good result in liver segmentation, and so can be considered a contribution to region growing algorithm used in this project.