Uma abordagem radiomics usando índices de diversidade filogenética e funcional para classificar nódulos de câncer de pulmão de células não pequenas em imagens de tomografia computadorizada

• Ano de defesa: 2019
• Autor(a) principal: Santos Neto, Antonino Calisto dos
• Orientador(a): Silva, Aristófanes Corrêa
• Banca de defesa: Silva, Aristófanes Corrêa , Cavalcante, André Borges , Carvalho Filho, Antônio Oséas de , Barros Netto, Stelmo Magalhaes
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal do Maranhão
• Programa de Pós-Graduação: PROGRAMA DE PÓS-GRADUAÇÃO EM ENGENHARIA DE ELETRICIDADE/CCET
• Departamento: COORDENAÇÃO DO CURSO DE ENGENHARIA ELÉTRICA/CCET
• País: Brasil
• Palavras-chave em Português: Radiomics; NSCLS-Radiomics; NSCLS; Índice diversidade filogenética; Índice de diversidade funcional; Classificação de nódulos pulmonares
• Palavras-chave em Inglês: Radiomics; NSCLS-Radiomics; NSCLS; Index of phylogenetic diversity; Functional diversity index; Classification of pulmonary nodules
• Área do conhecimento CNPq: Cancerologia; Metodologia e Técnicas da Computação
• Link de acesso: https://tedebc.ufma.br/jspui/handle/tede/2574

Lung cancer is the world’s largest cause of cancer death, accounting for more than 17% of all cancer-related deaths, with Non- Small Cell Lung Cancer (NSCLC) corresponds to approximately 85% of lung cancer occurrences. However, its early diagnosis may help in a sharp fall in this mortality rate. Due to the arduous process in the analysis of the imaging tests, an emerging field in image processing called Radiomics arises. This approach allows quantitative characterization of an image, which allows a much more precise definition of the tumor phenotype, using image processing and pattern recognition techniques, providing an early diagnosis of NSCLC quickly and helping the opinion of the specialist. Therefore, this work proposes a methodology for the classification of NSCLC nodules in Computed Tomography (CT) examinations using indexes of phylogenetic and functional diversity in a Radiomics approach. Divided into six steps, this methodology starts with the acquisition of NSCLC nodal images from the public NSCLC-Radiomics imaging base. In the second step, the lesions were extracted using the markings of the specialists. Then, in the third stage, quantizations are made to create greater diversity of species. In the fourth phase, texture characteristics are extracted based on phylogenetic and functional diversity indexes. Then, in the fifth phase, the characteristics are submitted to the classifications Suport Vector Machine, Random Forest and Random Tree. Finally, in the sixth step, the proposed methodology is validated using the area under the Receiver Operating Characteristic (ROC) curve, the Kappa index and the accuracy. The best values found for the classification of NSCLC nodules in the Radiomics approach resulted in a Kappa index of 0.990, an area under the ROC curve of 0.999 and an accuracy of 99.44.