Percolação multidimensional e multiescala para quantificação de imagens histológicas de linfomas
Carregando...
Data
2017-07-14
Autores
Roberto, Guilherme Freire [UNESP]
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
Neste trabalho é apresentado um novo método para quantificar e classificar amostras de tecido de linfomas não-Hodgkin, baseado na teoria da percolação. O método consiste em associar abordagens multiescala e multidimensional para dividir a imagem em regiões menores e verificar a similaridade de cor entre pixels. Um algoritmo de rotulagem de aglomerados é aplicado em cada região de interesse para obter os valores de números de aglomerados, ocorrência de percolação e ocupação do maior aglomerado. O método foi testado em diferentes classificadores com o objetivo de diferenciar três diferentes grupos de linfomas não-Hodgkin. Os resultados obtidos (taxas de AUC entre 0,940 e 0,993) superaram os fornecidos por métodos consolidados na literatura e permitiram identificar as três classes de linfomas não-Hodgkin: linfoma de células do manto, linfoma folicular e leucemia linfoide crônica.
In this work a new method to quantify and classify non-Hodgkin lymphomas tissue samples is presented, based upon percolation theory. The method consists of associating multiscale and multidimensional approaches in order to divide the image into smaller regions and then verifying color similarity between pixels. A cluster labeling algorithm was applied on each region of interest to obtain the values for the number of clusters, occurrence of percolation and occuppation of the biggest cluster. The method was tested on different classifiers aiming to differentiate three different groups of nonHodgkin lymphomas. The obtained results (AUC rates between 0.940 and 0.993) were compared to those provided by methods consolidated in the Literature, which indicates that percolation theory is a suitable approach for identifying these three classes of non-Hodgkin lymphomas, those being: mantle cell lymphoma, follicular lymphoma and chronic lymphocytic leukemia.
In this work a new method to quantify and classify non-Hodgkin lymphomas tissue samples is presented, based upon percolation theory. The method consists of associating multiscale and multidimensional approaches in order to divide the image into smaller regions and then verifying color similarity between pixels. A cluster labeling algorithm was applied on each region of interest to obtain the values for the number of clusters, occurrence of percolation and occuppation of the biggest cluster. The method was tested on different classifiers aiming to differentiate three different groups of nonHodgkin lymphomas. The obtained results (AUC rates between 0.940 and 0.993) were compared to those provided by methods consolidated in the Literature, which indicates that percolation theory is a suitable approach for identifying these three classes of non-Hodgkin lymphomas, those being: mantle cell lymphoma, follicular lymphoma and chronic lymphocytic leukemia.
Descrição
Palavras-chave
Atributos, Percolação, Multiescala, Multidimensional, Linfomas, Features, Percolation, Multiscale, Lymphomas