| Campo DC | Valor | Idioma |
|---|
| dc.contributor.advisor | Mendes, Cristiano Jacques Miosso Rodrigues | - |
| dc.contributor.author | Silva, Gabriela Barbosa | - |
| dc.date.accessioned | 2025-12-04T14:04:21Z | - |
| dc.date.available | 2025-12-04T14:04:21Z | - |
| dc.date.issued | 2025-12-04 | - |
| dc.date.submitted | 2025-01-08 | - |
| dc.identifier.citation | SILVA, Gabriela Barbosa. Simultaneous positron emission tomography and magnetic resonance image reconstruction with compressive sensing using cross-prior information and pre-filtering. 2025. 105 f., il. Dissertação (Mestrado em Engenharia Biomédica) — Universidade de Brasília, Brasília, 2025. | pt_BR |
| dc.identifier.uri | http://repositorio.unb.br/handle/10482/53340 | - |
| dc.description | Dissertação (Mestrado em Engenharia Biomédica) — Universidade de Brasília, Faculdade UnB Gama, Programa de pós-graduação em Engenharia Biomédica, 2025. | pt_BR |
| dc.description.abstract | A tecnologia de tomografia por emissão de positrons combinada com imagem de res sonância magnética (PET/MRI) constitui uma modalidade de imagem relativamente nova
que combina a capacidade do PET de rastrear o metabolismo com a resoluçao espacial da
Ressonância Magnética (MR).
Esta pesquisa utiliza informação a priori (IP) extraída de dados de imagens de PET, que
foram adquiridas simultaneamente com a RM do equipamento da Siemens® de PET/MRI,
Biograph mMR. E explorado o fato de que as imagens de PET e de RM são adquiridas
simultaneamente para usar informaçães da modalidade de PET com Compressive Sensing
(CS) para melhorar a reconstruçao das imagens de RM.
A metodologia foi testada variando o percentual de coeficientes da representação esparsa
considerada como posições de elementos nao nulos (e, portanto, usada como IP) de 0 a 90 %,
com 10 valores igualmente espaçados e variando o numero de ângulos usados na trajetéria
radial da ressonância magnética de 10 a 160 angulos, com 7 valores igualmente espaçados.
A qualidade das imagens foi avaliada quantitativamente, utilizando as métricas de signal to-error-ratio (SER) e Structural Similarity Index Measure (SSIM) para avaliar a qualidade
das imagens. As opiniães de especialistas em radiologia não foram coletadas para sustentar
uma analise qualitativa. Com uma proporção de IP de 10 %, a qualidade das imagens
reconstruédas melhora em comparação com aquelas reconstruédas sem IP para todas as
faixas de ângulos testados. | pt_BR |
| dc.description.sponsorship | Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES). | pt_BR |
| dc.language.iso | eng | pt_BR |
| dc.rights | Acesso Aberto | pt_BR |
| dc.title | Simultaneous positron emission tomography and magnetic resonance image reconstruction with compressive sensing using cross-prior information and pre-filtering | pt_BR |
| dc.title.alternative | Imagens de tomografia por emissão de pósitrons e ressonância magnética simultâneas com compressive sensing usando informação a priori cruzada e pré-filtragem | pt_BR |
| dc.type | Dissertação | pt_BR |
| dc.subject.keyword | Tomografia por emissão de pósitrons | pt_BR |
| dc.subject.keyword | Imagem por ressonância magnética | pt_BR |
| dc.subject.keyword | Compressive sensing | pt_BR |
| dc.subject.keyword | Informação a priori | pt_BR |
| dc.subject.keyword | Reconstrução de imagens | pt_BR |
| dc.rights.license | A 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. | pt_BR |
| dc.description.abstract1 | The technology of positron emission tomography combined with magnetic resonance
image (PET/MRI) constitutes a comparatively novel imaging modality that combines
PET’s ability to trace metabolism with MRI’s ability of tissue differentiation. This tool
analyzes the structure and function of tissues and organs.
Although PET/MRI offers many significant advantages, its limitations must be considered. These scans may take longer than MRI and PET/CT due to the need to acquire
detailed images from both modalities. This can increase the risk of movement during the
exam, affecting image quality.
In addition to artifacts caused by patient movements, the hardware of both scanners
interferes with each other, causing attenuation in the images. Furthermore, PET/MRI
technology is expensive to acquire and operate. The high cost of the scanners can be a
significant barrier, especially in health systems with limited resources.
This research uses prior information (PI) extracted from PET data, which was acquired simultaneously with MRI from Siemens ® PET-MR equipment, Biograph mMR.
We have explored the fact that PET and MR images are acquired simultaneously to use
information from PET modality to MRI and Compressive Sensing to enhance PET-MR
image reconstruction.
This methodology exploits PET information to reduce the number of samples required to obtain the same MR image quality obtained with more measurements. In CS
theory, this is equivalent to improving the quality of MR images for the same number of
measurements.
The proposed method extracts information corresponding to the positions of non-null
elements in the sparse representation of PET images, given that CS explores the sparse
representation. This requires aligning the PET and MR images, as well as adjusting the
spatial resolution to the same level, by nearest neighbor interpolation methods to reduce
the size of the images and bilinear interpolation to increase the size of the images.
We tested the approach by varying the percentage of coefficients in the sparse representation considered as non-null element positions (and therefore used as a priori information) from 0 to 90 %, with 10 equally spaced values. We also varied the number of
angles used from MRI from 10 to 160 angles, with 7 equally spaced values.
Image quality was evaluated quantitatively using signal-to-error-ratio (SER) and
structural similarity index measure (SSIM). The opinions of radiologist specialists were
not collected to sustain a qualitative analysis.With a PI proportion of 10 %, the reconstructed images quality improves compared to
those reconstructed without PI for all ranges of angles tested. Images reconstructed with
60 angles had an average SER of approximately 16dB and an average SSIM of roughly
0.82, while the images reconstructed with 10 % of PI had an average SER of 20dB and
an average SSIM of approximately 0.89.
Based on the results obtained in this study, the proposed method demonstrates the
potential to improve MRI image quality using PI from PET images extracted simultaneously from the same slice and spatial resolution. The use of PI from PET images for MRI
reconstruction increased image quality even with a low number of measurements. With
35 angles, we obtained a SER of 20 dB. This approach can help to reduce the number of
measurements required for MRI reconstruction in PET/MRI equipment.
For future work, we suggest the use of PI from MR image to reconstruct PET images. In addition, investigating this methodology for different trajectories and regions
of the body can provide relevant information about its applicability. Another potential
investigation is to use the k-space measurements extracted from the PET/MRI machine. | pt_BR |
| dc.description.unidade | Faculdade de Ciências e Tecnologias em Engenharia (FCTE) – Campus UnB Gama | pt_BR |
| dc.description.ppg | Programa de Pós-Graduação em Engenharia Biomédica | pt_BR |
| Aparece nas coleções: | Teses, dissertações e produtos pós-doutorado
|
