Deep learning for the diagnosis of Alzheimer's disease with 18F-FDG PET neuroimaging
Autor(a) principal: | |
---|---|
Data de Publicação: | 2021 |
Tipo de documento: | Dissertação |
Idioma: | eng |
Título da fonte: | Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) |
Texto Completo: | http://hdl.handle.net/10773/33725 |
Resumo: | Neurodegenerative disease is the term used for a range of incurable and debilitating conditions affecting the human's nervous system. Amongst these conditions, Alzheimer's Disease (AD) is responsible for the greatest burden both for the number of people affected and for the high costs in medical care. The challenges of the disease are related to the subtle symptoms, the increasing pace of disability and the long period of time over which patients will require special care. Recent research efforts have been dedicated to the development of computational tools that can be integrated into the workflow of doctors as a complement to support early diagnosis and targeted treatments. This dissertation aims to study the application of Deep Learning (DL) techniques for the automated classification of AD. The study focuses on the role of PET neuroimaging as a biomarker of neurodegenerative diseases, namely in classifying healthy versus AD patients. PET images of the cerebral metabolism of glucose with fluorine 18 (18F) fluorodeoxyglucose (18F FGD) were obtained from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database. The pre-processed dataset is used to train two Convolutional Neural Networks (CNNs). The first CNN architecture aims to explore transfer learning as a promising solution to the data challenge by using a 2D Inception V3 model, from Google, previously trained on a large dataset. This approach requires a preprocessing step in which the PET volumetric data is converted into a two-dimensional input image which is the input to the pre-trained model. The second approach involves a custom 3D-CNN to take advantage of spatial patterns on the full PET volumes by using 3D filters and 3D pooling layers. The comparative study highlights the performance and robustness of these two models in dealing with the limited availability of the labelled data. The performance of the estimators is evaluated through a cross-validation procedure, giving a score of 83.62% for the 2D-CNN and 86.80% for the 3D-CNN. The results achieved contribute to the understanding of the effectiveness of these methods in the diagnosis of AD. Given the expected margin for improvements, they can be considered promising and in line with the current state of the art. |
id |
RCAP_712b4cd12898c7508d80dc92d20a2fb5 |
---|---|
oai_identifier_str |
oai:ria.ua.pt:10773/33725 |
network_acronym_str |
RCAP |
network_name_str |
Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) |
repository_id_str |
7160 |
spelling |
Deep learning for the diagnosis of Alzheimer's disease with 18F-FDG PET neuroimagingAlzheimer's diseaseFDG-PET neuroimagingADNI datasetConvolutional neural networksTransfer learningCustom deep learningNeurodegenerative disease is the term used for a range of incurable and debilitating conditions affecting the human's nervous system. Amongst these conditions, Alzheimer's Disease (AD) is responsible for the greatest burden both for the number of people affected and for the high costs in medical care. The challenges of the disease are related to the subtle symptoms, the increasing pace of disability and the long period of time over which patients will require special care. Recent research efforts have been dedicated to the development of computational tools that can be integrated into the workflow of doctors as a complement to support early diagnosis and targeted treatments. This dissertation aims to study the application of Deep Learning (DL) techniques for the automated classification of AD. The study focuses on the role of PET neuroimaging as a biomarker of neurodegenerative diseases, namely in classifying healthy versus AD patients. PET images of the cerebral metabolism of glucose with fluorine 18 (18F) fluorodeoxyglucose (18F FGD) were obtained from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database. The pre-processed dataset is used to train two Convolutional Neural Networks (CNNs). The first CNN architecture aims to explore transfer learning as a promising solution to the data challenge by using a 2D Inception V3 model, from Google, previously trained on a large dataset. This approach requires a preprocessing step in which the PET volumetric data is converted into a two-dimensional input image which is the input to the pre-trained model. The second approach involves a custom 3D-CNN to take advantage of spatial patterns on the full PET volumes by using 3D filters and 3D pooling layers. The comparative study highlights the performance and robustness of these two models in dealing with the limited availability of the labelled data. The performance of the estimators is evaluated through a cross-validation procedure, giving a score of 83.62% for the 2D-CNN and 86.80% for the 3D-CNN. The results achieved contribute to the understanding of the effectiveness of these methods in the diagnosis of AD. Given the expected margin for improvements, they can be considered promising and in line with the current state of the art.Doença neurodegenerativa é um termo utilizado para uma série de condições incuráveis e debilitantes que afetam o sistema nervoso humano. Destas condições, a doença de Alzheimer (DA) é a mais preocupante, tanto pelo número de pessoas afetadas como pelos elevados custos em tratamento medico. Os principais desafios associados a esta doença estão relacionados com os sintomas subtis, o rápido desenvolvimento de incapacidade e ao longo período de tempo durante o qual os pacientes necessitarão de cuidados especiais. Pesquisas recentes têm sido dedicadas ao desenvolvimento de ferramentas computacionais capazes de ser integradas nos procedimentos médicos como complemento para apoiar o diagnóstico precoce e tratamentos adequados. Esta dissertação procura estudar a aplicação de técnicas de aprendizagem profunda (AP) na classificação automatizada da DA. Este estudo tem como foco principal o papel da neuroimagem PET como biomarcador de doenças neurodegenerativas, especialmente na classificação de pacientes saudáveis em comparação com pacientes com DA. Imagens PET do metabolismo cerebral de glucose com flúor-18 (18F) fluorodesoxiglucose (18F FGD) foram obtidas através da base de dados da Alzheimer's Dissesse Neuroimaging Initiative (ADNI). O dataset pré-processado é usado para treinar duas redes neurais convulsionais (RNCs). A arquitetura da primeira RNC procura explorar a transferência de aprendizagem como uma solução promissora para o problema dos dados através da utilização de um modelo Inception V3 2D, da Google, previamente treinado num dataset maior. Esta abordagem requer um passo de pré -processamento onde dados volumétricos PET são convertidos numa imagem bidimensional que por sua vez será os dados de entrada do modelo pré-treinado. A segunda abordagem involve uma RNC 3D personalizada de maneira a utilizar os padrões espaciais presentes nos volumes PET através de filtros 3D e camadas de pooling 3D. O estudo comparativo foca-se no desempenho e robustez dos dois modelos ao lidar com a disponibilidade limitada de dados classificados. O desempenho dos classificadores é avaliado através de um processo de validação cruzada, atribuindo uma pontuação de 83.62% à RNC 2D e de 86.80% à RNC 3D. Os resultados obtidos contribuem para análise da eficácia destes métodos no diagnóstico da DA. Tendo em conta as melhorias expectáveis, estas poderam ser consideradas abordagens promissoras e de acordo com o atual estado da arte.2022-04-26T07:56:05Z2021-12-17T00:00:00Z2021-12-17info:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/masterThesisapplication/pdfhttp://hdl.handle.net/10773/33725engBastos, José Carlos Alho Barros deinfo:eu-repo/semantics/openAccessreponame:Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)instname:Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informaçãoinstacron:RCAAP2024-02-22T12:04:51Zoai:ria.ua.pt:10773/33725Portal AgregadorONGhttps://www.rcaap.pt/oai/openaireopendoar:71602024-03-20T03:05:04.645388Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) - Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informaçãofalse |
dc.title.none.fl_str_mv |
Deep learning for the diagnosis of Alzheimer's disease with 18F-FDG PET neuroimaging |
title |
Deep learning for the diagnosis of Alzheimer's disease with 18F-FDG PET neuroimaging |
spellingShingle |
Deep learning for the diagnosis of Alzheimer's disease with 18F-FDG PET neuroimaging Bastos, José Carlos Alho Barros de Alzheimer's disease FDG-PET neuroimaging ADNI dataset Convolutional neural networks Transfer learning Custom deep learning |
title_short |
Deep learning for the diagnosis of Alzheimer's disease with 18F-FDG PET neuroimaging |
title_full |
Deep learning for the diagnosis of Alzheimer's disease with 18F-FDG PET neuroimaging |
title_fullStr |
Deep learning for the diagnosis of Alzheimer's disease with 18F-FDG PET neuroimaging |
title_full_unstemmed |
Deep learning for the diagnosis of Alzheimer's disease with 18F-FDG PET neuroimaging |
title_sort |
Deep learning for the diagnosis of Alzheimer's disease with 18F-FDG PET neuroimaging |
author |
Bastos, José Carlos Alho Barros de |
author_facet |
Bastos, José Carlos Alho Barros de |
author_role |
author |
dc.contributor.author.fl_str_mv |
Bastos, José Carlos Alho Barros de |
dc.subject.por.fl_str_mv |
Alzheimer's disease FDG-PET neuroimaging ADNI dataset Convolutional neural networks Transfer learning Custom deep learning |
topic |
Alzheimer's disease FDG-PET neuroimaging ADNI dataset Convolutional neural networks Transfer learning Custom deep learning |
description |
Neurodegenerative disease is the term used for a range of incurable and debilitating conditions affecting the human's nervous system. Amongst these conditions, Alzheimer's Disease (AD) is responsible for the greatest burden both for the number of people affected and for the high costs in medical care. The challenges of the disease are related to the subtle symptoms, the increasing pace of disability and the long period of time over which patients will require special care. Recent research efforts have been dedicated to the development of computational tools that can be integrated into the workflow of doctors as a complement to support early diagnosis and targeted treatments. This dissertation aims to study the application of Deep Learning (DL) techniques for the automated classification of AD. The study focuses on the role of PET neuroimaging as a biomarker of neurodegenerative diseases, namely in classifying healthy versus AD patients. PET images of the cerebral metabolism of glucose with fluorine 18 (18F) fluorodeoxyglucose (18F FGD) were obtained from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database. The pre-processed dataset is used to train two Convolutional Neural Networks (CNNs). The first CNN architecture aims to explore transfer learning as a promising solution to the data challenge by using a 2D Inception V3 model, from Google, previously trained on a large dataset. This approach requires a preprocessing step in which the PET volumetric data is converted into a two-dimensional input image which is the input to the pre-trained model. The second approach involves a custom 3D-CNN to take advantage of spatial patterns on the full PET volumes by using 3D filters and 3D pooling layers. The comparative study highlights the performance and robustness of these two models in dealing with the limited availability of the labelled data. The performance of the estimators is evaluated through a cross-validation procedure, giving a score of 83.62% for the 2D-CNN and 86.80% for the 3D-CNN. The results achieved contribute to the understanding of the effectiveness of these methods in the diagnosis of AD. Given the expected margin for improvements, they can be considered promising and in line with the current state of the art. |
publishDate |
2021 |
dc.date.none.fl_str_mv |
2021-12-17T00:00:00Z 2021-12-17 2022-04-26T07:56:05Z |
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/10773/33725 |
url |
http://hdl.handle.net/10773/33725 |
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:Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) instname:Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informação instacron:RCAAP |
instname_str |
Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informação |
instacron_str |
RCAAP |
institution |
RCAAP |
reponame_str |
Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) |
collection |
Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) |
repository.name.fl_str_mv |
Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) - Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informação |
repository.mail.fl_str_mv |
|
_version_ |
1799137705755410432 |