The application of analysis filters in compressed sensing algorithms for magnetic resonance imaging reconstruction

Registro completo de metadados
MetadadosDescriçãoIdioma
Autor(es): dc.contributorFarias, Mylène Christine Queiroz de-
Autor(es): dc.contributorMiosso, Cristiano Jacques-
Autor(es): dc.contributorjonathanalis@gmail.com-
Autor(es): dc.creatorLima, Jonathan Alis Salgado-
Data de aceite: dc.date.accessioned2024-10-23T15:02:22Z-
Data de disponibilização: dc.date.available2024-10-23T15:02:22Z-
Data de envio: dc.date.issued2020-06-15-
Data de envio: dc.date.issued2020-06-15-
Data de envio: dc.date.issued2020-06-15-
Data de envio: dc.date.issued2019-08-09-
Fonte completa do material: dc.identifierhttps://repositorio.unb.br/handle/10482/38031-
Fonte: dc.identifier.urihttp://educapes.capes.gov.br/handle/capes/872724-
Descrição: dc.descriptionTese (doutorado)—Universidade de Brasília, Instituto de Ciências Exatas, Departamento de Ciência da Computação, 2019.-
Descrição: dc.descriptionCoordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES).-
Descrição: dc.descriptionMagnetic Resonance Imaging (MRI) exams usually take a long time to be performed because they require a great amount of measurements to reconstruct an image with good quality. Decreasing the acquisition time of MRI can prevent motion artifacts, make possible to perform new types of exams, and also reduce MRI costs. Compressed Sensing (CS) techniques are able to reconstruct MRI images at a sub- Nyquist rate, provided that the signals are sparse in a known domain. A CS method known as total variation (TV) minimization, minimizes the nite di erences to reconstruct the signal. This operation can be interpreted as a ltering operation that is performed in the reconstruction steps. On the other hand, the pre- ltering method reconstructs ltered versions of the image with CS and recombine their spectrum to obtain a better image quality. This method relies on the fact that (high-pass) ltered versions of the images are sparse in the pixel domain and can be reconstructed with CS using fewer measurements. In this work, I use ltering methods with CS to improve the quality of the undersampled MRI image reconstructions. The lters provide sparsity to the images, and generate better CS reconstructions. In the pre- ltering methods, I proposed a systematical test to evaluate a large number of lter banks, which were still not tested in the pre- ltering literature. I also proposed a threshold method to include measurements in the solution space, based on the stop-band of the lters. Finally, I proposed the ltering norms, a method that uses lters in the reconstruction algorithm. This method generalizes the TV minimization for any type of lter. I simulated the methods extensively for di erent sampling density and on a large set of images, and use objective metrics to evaluate the reconstruction quality. The pre- ltering, for low order lters designed with windowing method obtained SNR values between 1 and 2.9 dB higher than the TV minimization. Filtering norms with a combination of lters resulted in SNR values between 1.2 and 1.5 dB higher than values obtained with the TV. In most cases, the threshold method improved the image quality results. However, the highest quality improvements were observed for poor reconstructions.-
Descrição: dc.descriptionInstituto de Ciências Exatas (IE)-
Descrição: dc.descriptionDepartamento de Ciência da Computação (IE CIC)-
Descrição: dc.descriptionPrograma de Pós-Graduação em Informática-
Formato: dc.formatapplication/pdf-
Direitos: dc.rightsAcesso Aberto-
Direitos: dc.rightsA concessão da licença deste item refere-se ao termo de autorização impresso assinado pelo autor com as seguintes condições: Na qualidade de titular dos direitos de autor da publicação, autorizo a Universidade de Brasília e o IBICT a disponibilizar por meio dos sites www.bce.unb.br, www.ibict.br, http://hercules.vtls.com/cgi-bin/ndltd/chameleon?lng=pt&skin=ndltd sem ressarcimento dos direitos autorais, de acordo com a Lei nº 9610/98, o texto integral da obra disponibilizada, conforme permissões assinaladas, para fins de leitura, impressão e/ou download, a título de divulgação da produção científica brasileira, a partir desta data.-
Palavras-chave: dc.subjectCompressed sensing-
Palavras-chave: dc.subjectImageamento médico-
Palavras-chave: dc.subjectFiltragem-
Palavras-chave: dc.subjectRessonância magnética-
Título: dc.titleThe application of analysis filters in compressed sensing algorithms for magnetic resonance imaging reconstruction-
Título: dc.titleMétodos de filtragem digital em compressed sensing para reconstrução de imagens de ressonância magética-
Tipo de arquivo: dc.typelivro digital-
Aparece nas coleções:Repositório Institucional – UNB

Não existem arquivos associados a este item.