Convolutional Neural Networks in Prostate Cancer Detection, Segmentation and Classification using mpMRI images and feature-selected Radiomic Features

Bibliographic Details
Main Author: Fidalgo, Miguel Maria Santos
Publication Date: 2023
Format: Master thesis
Language: eng
Source: Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
Download full: http://hdl.handle.net/10451/62951
Summary: Tese de Mestrado, Engenharia Biomédica e Biofísica, 2024, Universidade de Lisboa, Faculdade de Ciências
id RCAP_828b46d050e9758c43fb8b02300d0b06
oai_identifier_str oai:repositorio.ulisboa.pt:10451/62951
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 Convolutional Neural Networks in Prostate Cancer Detection, Segmentation and Classification using mpMRI images and feature-selected Radiomic FeaturesPróstataCancroRessonância Magnética MultiparamétricaU-NetRadiomicsSeleção de featuresTese de mestrado 2024Domínio/Área Científica::Engenharia e Tecnologia::Engenharia MédicaTese de Mestrado, Engenharia Biomédica e Biofísica, 2024, Universidade de Lisboa, Faculdade de CiênciasIn many parts of the western world, prostate cancer is the most diagnosed non-cutaneous cancer in men [4]. Recently, Multi-Parametric Magnetic Resonance Imaging (mpMRI) has been ex plored as a tool for screening and evaluation of prostate cancer, alongside traditional techniques like the Prostate Specific Antigen (PSA) test and biopsy. Analysis of mpMRI requires experience, expertise and time. Machine learning and algorithm based image analysis can be used to assist the radiologist with an automated analysis of the images. For prostate cancer diagnosis and monitoring, these models and algorithms aim to segment the gland, its regions and possible lesions, while classifying them, according to systems like the Prostate Imaging Reporting Data System (PI-RADS) standard [5]. Radiomic features can be used to extract additional information from an image. These reflect various patterns and textures in the MRI image which can indicate abnormalities and prostate cancer. However, the high volume of radiomic features that can be extracted can be overwhelming, so determining which features are the most useful for cancer prediction can be a desirable study. In this dissertation, a procedure for selecting the best features is described. The workflow was designed for a high number of features and multidimensional data, like mpMRI and lesion segmentations. Through Recurdive Feature Elimination (RFE), an array of radiomic features was selected and used on a U-Net for prostate cancer prediction, in order to validate the whole process. The results selected the Wavelet-LLL as the best filter for radiomic feature extraction and the Emphasis features (texture features) as the best for prostate cancer prediction. The final U-Net model built for validation of the results of feature selection displayed acceptable performance overall, with good lesion segmentation capabilities (0,8831 True Positive Ratio (TPR) in the context of the Ground Truth (GT)). It had, however, some difficulties to classify said lesions correctly (71,64% of PI-RADS 4 lesions were misclassified as PI-RADS 5), showing a bias towards classifying most lesions as PI-RADS 5.Conceição, RaquelFinn, SéanRepositório da Universidade de LisboaFidalgo, Miguel Maria Santos202420232026-10-30T00:00:00Z2024-01-01T00:00:00Zinfo:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/masterThesisapplication/pdfhttp://hdl.handle.net/10451/62951TID:203881621enginfo:eu-repo/semantics/embargoedAccessreponame:Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)instname:FCCN, serviços digitais da FCT – Fundação para a Ciência e a Tecnologiainstacron:RCAAP2025-03-17T15:12:08Zoai:repositorio.ulisboa.pt:10451/62951Portal AgregadorONGhttps://www.rcaap.pt/oai/openaireinfo@rcaap.ptopendoar:https://opendoar.ac.uk/repository/71602025-05-29T03:36:30.871005Repositórios 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 Convolutional Neural Networks in Prostate Cancer Detection, Segmentation and Classification using mpMRI images and feature-selected Radiomic Features
title Convolutional Neural Networks in Prostate Cancer Detection, Segmentation and Classification using mpMRI images and feature-selected Radiomic Features
spellingShingle Convolutional Neural Networks in Prostate Cancer Detection, Segmentation and Classification using mpMRI images and feature-selected Radiomic Features
Fidalgo, Miguel Maria Santos
Próstata
Cancro
Ressonância Magnética Multiparamétrica
U-Net
Radiomics
Seleção de features
Tese de mestrado 2024
Domínio/Área Científica::Engenharia e Tecnologia::Engenharia Médica
title_short Convolutional Neural Networks in Prostate Cancer Detection, Segmentation and Classification using mpMRI images and feature-selected Radiomic Features
title_full Convolutional Neural Networks in Prostate Cancer Detection, Segmentation and Classification using mpMRI images and feature-selected Radiomic Features
title_fullStr Convolutional Neural Networks in Prostate Cancer Detection, Segmentation and Classification using mpMRI images and feature-selected Radiomic Features
title_full_unstemmed Convolutional Neural Networks in Prostate Cancer Detection, Segmentation and Classification using mpMRI images and feature-selected Radiomic Features
title_sort Convolutional Neural Networks in Prostate Cancer Detection, Segmentation and Classification using mpMRI images and feature-selected Radiomic Features
author Fidalgo, Miguel Maria Santos
author_facet Fidalgo, Miguel Maria Santos
author_role author
dc.contributor.none.fl_str_mv Conceição, Raquel
Finn, Séan
Repositório da Universidade de Lisboa
dc.contributor.author.fl_str_mv Fidalgo, Miguel Maria Santos
dc.subject.por.fl_str_mv Próstata
Cancro
Ressonância Magnética Multiparamétrica
U-Net
Radiomics
Seleção de features
Tese de mestrado 2024
Domínio/Área Científica::Engenharia e Tecnologia::Engenharia Médica
topic Próstata
Cancro
Ressonância Magnética Multiparamétrica
U-Net
Radiomics
Seleção de features
Tese de mestrado 2024
Domínio/Área Científica::Engenharia e Tecnologia::Engenharia Médica
description Tese de Mestrado, Engenharia Biomédica e Biofísica, 2024, Universidade de Lisboa, Faculdade de Ciências
publishDate 2023
dc.date.none.fl_str_mv 2023
2024
2024-01-01T00:00:00Z
2026-10-30T00:00:00Z
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/10451/62951
TID:203881621
url http://hdl.handle.net/10451/62951
identifier_str_mv TID:203881621
dc.language.iso.fl_str_mv eng
language eng
dc.rights.driver.fl_str_mv info:eu-repo/semantics/embargoedAccess
eu_rights_str_mv embargoedAccess
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_ 1833601762611691520

Similar Items