Banca de DEFESA: THAINA APARECIDA AZEVEDO TOSTA

Uma banca de DEFESA de DOUTORADO foi cadastrada pelo programa.
DISCENTE : THAINA APARECIDA AZEVEDO TOSTA
DATA : 20/01/2020
HORA: 14:00
LOCAL: na sala 306, 3º andar, Bloco B, Campus SA da Fundação Universidade Federal do ABC, localizada na Avenida dos Estados, 5001, Santa Terezinha, Santo André, SP
TÍTULO:

Computational Algorithms for Normalization of Hematoxylin and Eosin Stain in Digital Histological Images

PÁGINAS: 148
GRANDE ÁREA: Engenharias
ÁREA: Engenharia Biomédica
SUBÁREA: Bioengenharia
ESPECIALIDADE: Processamento de Sinais Biológicos
RESUMO:

Diagnoses of different types of cancer can be confirmed by microscopic analyses of tissue samples collected from the patient's body. With the digitization of these samples, it is possible to obtain the so-called histological images, which allow the development of computer systems that aid pathologists in the definition of diagnoses. However, these images are subject to color variations which can reduce the performance of computational methods of histological image processing. For correcting these variations, normalization techniques are used to adjust the colors of these images. In this work, new algorithms are proposed for normalization of histological images stained with hematoxylin and eosin, dyes commonly used in clinical practices. To do so, biological concepts and from the stain are used for the estimates of the stain color appearance matrices and stain density maps. The estimated matrices define the color of the dyes in the RGB color space, and the maps estimate the amount of each dye represented by the image's pixels. After performing these estimates for the original and reference images, the reference matrix is combined with the original map so the reference image colors are transferred to the original image, while the representation of its structures is preserved. This proposal was evaluated with histological images of different types of cancer, with clear color variations, demonstrating relevant results by the used evaluation criteria. Quantitative metrics indicated the good performance of the proposed method in comparison with techniques already published in the literature. Among these evaluations, it is worth highlighting the high performance of this proposal for the representation of the eosin, its advantageous application for the results of segmentation and classification of histological images, and its high computational efficiency.

MEMBROS DA BANCA:
Presidente - Interno ao Programa - 254.009.018-40 - MARCELO ZANCHETTA DO NASCIMENTO - UFU
Membro Titular - Examinador(a) Interno ao Programa - 1672975 - JOAO RICARDO SATO
Membro Titular - Examinador(a) Externo ao Programa - 1722875 - DAVID CORREA MARTINS JUNIOR
Membro Titular - Examinador(a) Externo à Instituição - MARCELO COSTA OLIVEIRA - UFAL
Membro Titular - Examinador(a) Externo à Instituição - HELIO PEDRINI - UNICAMP
Membro Suplente - Examinador(a) Interno ao Programa - 2566275 - GUSTAVO SOUSA PAVANI
Membro Suplente - Examinador(a) Externo à Instituição - MÁRCIA APARECIDA SILVA BISSACO - UMC
Membro Suplente - Examinador(a) Externo à Instituição - RODRIGO PEREIRA RAMOS - UNIVASF