Evaluation of image registration performance and study of classification algorithm on histopathological images

Detalhes bibliográficos
Autor(a) principal: Luzio, Jessica Monteiro
Data de Publicação: 2018
Tipo de documento: Dissertação
Idioma: eng
Título da fonte: Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
Texto Completo: http://hdl.handle.net/10451/36322
Resumo: Tese de mestrado integrado em Engenharia Biomédica e Biofísica, apresentada à Universidade de Lisboa, através da Faculdade de Ciências, em 2018
id RCAP_6da4ea8ff1a77649c9e3e539ec732245
oai_identifier_str oai:repositorio.ulisboa.pt:10451/36322
network_acronym_str RCAP
network_name_str Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
repository_id_str https://opendoar.ac.uk/repository/7160
spelling Evaluation of image registration performance and study of classification algorithm on histopathological imagesPatologia digitalHistopatologiaRegisto de imagemClassificação de imagemAnálise de imagemTeses de mestrado - 2018Domínio/Área Científica::Engenharia e Tecnologia::Engenharia MédicaTese de mestrado integrado em Engenharia Biomédica e Biofísica, apresentada à Universidade de Lisboa, através da Faculdade de Ciências, em 2018The project described in this thesis was developed in the Roche Pharma Research and Early Development (pRED) department in Penzberg, which is focused on development of personalized health care strategies. The primary goal of the study is to explore computational tools used in Digital Pathology, more specifically on image registration and classification. Digital Pathology refers to animage-based environment that enables the conversion of glass slides into digital images, and also the management and interpretation of it, involving an enormous amount of data to be handled. The image analysis tasks are fundamental since they allow the identification and classification of different types of structures on tissue samples. Staining techniques as Hematoxylin and Eosin (H&E) and Immunohistochemistr (IHC) techniques are used, allowing the extraction of relevant information for the study of pathological processes. Image registration tools to transfer the annotations from one slide to other are important and routinely used in image analysis processes. Image classification is also a key task to identify and distinguish different type of structures in the tissue. The first part of this project investigates the feasibility of an automated method for image registration assessment whereas the second part studies the influence of classifiers, features and training set changes in the classification performance. With this we aim to gather information for further improvements on the process of image analysis. After applied the registration algorithm, a distance based method was developed to evaluate its performance. Distances between each point on target and source annotations were calculated and summed up in the median value. The results of automatic annotations were compared with the manual, in order to ascertain if they behave the same way and so, if the automatic approach can be used for performance evaluation. The assumption that the closest points between two annotations are the correspondent tissue points, makes the results to not give valid information when the registration is not good. As the comparison is done between tissue outline, even when the two annotations are misaligned, there are still close points between the two, even though they do not correspond to the same tissue parts. This approach is then not reliable to use for performance assessment. Regarding image classification, an algorithm for vessels detection is used and three different settings are tested in order to identify the influence on classification performance: classifiers, features and training set. For the classifiers no significant differences were identified. It was also verified that increasing the number of features does not necessarily increase the classification performance. With regard to the training set, for one of the ground truth provided, a good improvement was verified when balancing the data set. However, compared with information provided by a different pathologist, the results were not improved with the balancing. This can also be due to the under sampling done for balancing, which might have discarded useful samples. Although this study may not provide sufficient information to draw general conclusions, it is a starting point for future studies that may improve what was done so far.Matela, Nuno Miguel de Pinto Lobo e,1978-Grimm, OliverRepositório da Universidade de LisboaLuzio, Jessica Monteiro2019-01-09T19:07:37Z201820182018-01-01T00:00:00Zinfo:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/masterThesisapplication/pdfhttp://hdl.handle.net/10451/36322TID:202184684enginfo:eu-repo/semantics/openAccessreponame:Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)instname:FCCN, serviços digitais da FCT – Fundação para a Ciência e a Tecnologiainstacron:RCAAP2025-03-17T14:02:00Zoai:repositorio.ulisboa.pt:10451/36322Portal AgregadorONGhttps://www.rcaap.pt/oai/openaireinfo@rcaap.ptopendoar:https://opendoar.ac.uk/repository/71602025-05-29T03:01:00.553257Repositórios Científicos de Acesso Aberto de Portugal (RCAAP) - FCCN, serviços digitais da FCT – Fundação para a Ciência e a Tecnologiafalse
dc.title.none.fl_str_mv Evaluation of image registration performance and study of classification algorithm on histopathological images
title Evaluation of image registration performance and study of classification algorithm on histopathological images
spellingShingle Evaluation of image registration performance and study of classification algorithm on histopathological images
Luzio, Jessica Monteiro
Patologia digital
Histopatologia
Registo de imagem
Classificação de imagem
Análise de imagem
Teses de mestrado - 2018
Domínio/Área Científica::Engenharia e Tecnologia::Engenharia Médica
title_short Evaluation of image registration performance and study of classification algorithm on histopathological images
title_full Evaluation of image registration performance and study of classification algorithm on histopathological images
title_fullStr Evaluation of image registration performance and study of classification algorithm on histopathological images
title_full_unstemmed Evaluation of image registration performance and study of classification algorithm on histopathological images
title_sort Evaluation of image registration performance and study of classification algorithm on histopathological images
author Luzio, Jessica Monteiro
author_facet Luzio, Jessica Monteiro
author_role author
dc.contributor.none.fl_str_mv Matela, Nuno Miguel de Pinto Lobo e,1978-
Grimm, Oliver
Repositório da Universidade de Lisboa
dc.contributor.author.fl_str_mv Luzio, Jessica Monteiro
dc.subject.por.fl_str_mv Patologia digital
Histopatologia
Registo de imagem
Classificação de imagem
Análise de imagem
Teses de mestrado - 2018
Domínio/Área Científica::Engenharia e Tecnologia::Engenharia Médica
topic Patologia digital
Histopatologia
Registo de imagem
Classificação de imagem
Análise de imagem
Teses de mestrado - 2018
Domínio/Área Científica::Engenharia e Tecnologia::Engenharia Médica
description Tese de mestrado integrado em Engenharia Biomédica e Biofísica, apresentada à Universidade de Lisboa, através da Faculdade de Ciências, em 2018
publishDate 2018
dc.date.none.fl_str_mv 2018
2018
2018-01-01T00:00:00Z
2019-01-09T19:07:37Z
dc.type.status.fl_str_mv info:eu-repo/semantics/publishedVersion
dc.type.driver.fl_str_mv info:eu-repo/semantics/masterThesis
format masterThesis
status_str publishedVersion
dc.identifier.uri.fl_str_mv http://hdl.handle.net/10451/36322
TID:202184684
url http://hdl.handle.net/10451/36322
identifier_str_mv TID:202184684
dc.language.iso.fl_str_mv eng
language eng
dc.rights.driver.fl_str_mv info:eu-repo/semantics/openAccess
eu_rights_str_mv openAccess
dc.format.none.fl_str_mv application/pdf
dc.source.none.fl_str_mv reponame:Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
instname:FCCN, serviços digitais da FCT – Fundação para a Ciência e a Tecnologia
instacron:RCAAP
instname_str FCCN, serviços digitais da FCT – Fundação para a Ciência e a Tecnologia
instacron_str RCAAP
institution RCAAP
reponame_str Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
collection Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
repository.name.fl_str_mv Repositórios Científicos de Acesso Aberto de Portugal (RCAAP) - FCCN, serviços digitais da FCT – Fundação para a Ciência e a Tecnologia
repository.mail.fl_str_mv info@rcaap.pt
_version_ 1833601562583236608

Registros relacionados