Publicado em: 29/04/2013

Dissertação de Mestrado em Processamento de Imagens

UNIVERSIDADE FEDERAL DO RIO GRANDE DO SUL
INSTITUTO DE INFORMÁTICA
PROGRAMA DE POS-GRADUAÇÃO EM COMPUTAÇÃO
———————————————————
DEFESA DE DISSERTAÇÃO DE MESTRADO

Aluno: Pablo Gautério Cavalcanti
Orientador: Prof. Dr. Jacob Scharcanski

Título: Automated Prescreening of Melanocytic Skin Lesions Using Standard Camera Images
Linha de Pesquisa: Processamento de Imagens

Data: 02/05/2013
Hora: 14h
Local: Sala 215 Prédio 43412 – Instituto de Informática

Banca Examinadora:
Prof. Dr. M. Emre Celebi (LSUS – EUA) Por videoconferência
Prof. Dr. Paul Douglas Fisher (UFRGS – Medicina)
Prof. Dr. Valter Roesler (UFRGS)

Presidente da Banca: Prof. Dr. Jacob Scharcanski

Resumo:

Melanoma is a type of malignant pigmented skin lesion, and currently is among the most dangerous existing cancers. However, differentiating malignant and benign cases is a hard task even for experienced specialists, and a computer-aided diagnosis system can be an useful tool. Usually, the system starts by pre-processing the image, i.e. removing undesired artifacts such as hair, freckles or shading effects. Next, the system performs a segmentation step to identify the lesion boundaries. Finally, based on the image area identified as lesion, several features are computed and a classification is provided. In this Thesis, presented as a collection of published papers, we detail approaches to automatically execute all these steps, resulting in a pre-diagnosis for a pigmented skin lesion based only in a standard camera image (i.e. a simple color photograph). We tested our methods on publicly available datasets and achieved better segmentation and classification results than methods previously proposed in the literature.

Palavras-Chave: Palavras-chave: Melanocytic skin lesions classification, Image processing, Melanoma.