prikaz prve stranice dokumenta Unaprijeđivanje kvalitete CT snimaka primjenom neuronskih mreža
  Download
PDF 9 MB
master's thesis
Unaprijeđivanje kvalitete CT snimaka primjenom neuronskih mreža
2019. urn:nbn:hr:200:848536
Pavlović, Kristijan
Josip Juraj Strossmayer University of Osijek
Faculty of Electrical Engineering, Computer Science and Information Technology Osijek
Department of Software Engineering
Chair of Visual Computing

Cite this document

Pavlović, K. (2019). Unaprijeđivanje kvalitete CT snimaka primjenom neuronskih mreža (Master's thesis). Retrieved from https://urn.nsk.hr/urn:nbn:hr:200:848536

Pavlović, Kristijan. "Unaprijeđivanje kvalitete CT snimaka primjenom neuronskih mreža." Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, 2019. https://urn.nsk.hr/urn:nbn:hr:200:848536

Pavlović, Kristijan. "Unaprijeđivanje kvalitete CT snimaka primjenom neuronskih mreža." Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, 2019. https://urn.nsk.hr/urn:nbn:hr:200:848536

Pavlović, K. (2019). 'Unaprijeđivanje kvalitete CT snimaka primjenom neuronskih mreža', Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, accessed 28 July 2021, https://urn.nsk.hr/urn:nbn:hr:200:848536

Pavlović K. Unaprijeđivanje kvalitete CT snimaka primjenom neuronskih mreža [Master's thesis]. Osijek: Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek; 2019 [cited 2021 July 28] Available at: https://urn.nsk.hr/urn:nbn:hr:200:848536

K. Pavlović, "Unaprijeđivanje kvalitete CT snimaka primjenom neuronskih mreža", Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, Osijek, 2019. Available at: https://urn.nsk.hr/urn:nbn:hr:200:848536

Islandora Bookmark - Please login to use this feature
Metadata
TitleUnaprijeđivanje kvalitete CT snimaka primjenom neuronskih mreža
Title (english)Neural network-based CT image quality enhancement
AuthorKristijan Pavlović
MentorIrena Galić (mentor)
MentorHrvoje Leventić (mentor)
Committee memberIrena Galić (predsjednik povjerenstva)
Committee memberČaslav Livada (član povjerenstva)
Committee memberKrešimir Romić (član povjerenstva)
GranterJosip Juraj Strossmayer University of Osijek
Faculty of Electrical Engineering, Computer Science and Information Technology Osijek
(Department of Software Engineering)
(Chair of Visual Computing)
Osijek
Defense date and country2019-09-21, Croatia
Scientific / art field,
discipline and subdiscipline		TECHNICAL SCIENCES
Computing
Artificial Intelligence
Abstract
Iako su CT snimke pogodne za prikaz koštanih struktura, slobodnih zglobnih tijela te topografskog odnosa koštanih i mekotkivnih struktura, njihov rastu´ci broj pove´cava potencijalni rizik od zraˇcenja i postaje razlog zabrinutosti javnosti. Da bi se smanjio potencijalni rizik, CT s niskom dozom zraˇcenja privlaˇci sve ve´cu pozornost, ali smanjivanje koliˇcine zraˇcenja znatno pogoršava kvalitetu CT snimke. U sklopu ovog diplomskog rada prouˇcene su state of the art metode te su implementirane dvije konvolucijske neuronske mreže za unaprje ¯ divanje kvalitete CT snimki. Opisan je postupak predobrade kojim se skup podataka prilagodio svakoj od mreža, opisan je proces uˇcenja te su uspore ¯ deni rezultati njihove evaluacije. Budu´ci da je konvolucijska mreža za unaprje ¯ divanje kvalitete CT snimki koja se temelji na U-net arhitekturi ostvarila bolje rezultate u svakom aspektu evaluacije, predlaže se njeno korištenje za rješavanje zadanog problema. Usporedbom rezultata dobivenih kvantitativnom evaluacijom predložene mreže s rezultatima state of the art metoda može se zakljuˇciti kako im je predložena mreža konkurentna
Abstract (english)
Although CT scans are suitable for displaying bone structures, joints, and the topographic relationship of bone and soft tissue structures, their growing number increases the potential risk of radiation and is a cause for public concern. To reduce the potential risk, low-dose CT is gaining increasing attention, but reducing the amount of radiation significantly degrades the quality of the CT scan. As part of this graduate thesis, state of the art methods were studied and two convolutional neural networks were implemented to improve the quality of CT images. Both data preprocessing, which was used to adapt the dataset to each network, and the learning process were described, and the evaluation results of both networks were compared. As the convolutional network for CT image enhancement based on the U-net architecture has achieved better results in every aspect of the evaluation, its use is proposed as a solution for the given problem. Comparing the results obtained by quantitative evaluation of the proposed network with the results of the state of the art methods, one can conclude that the results closely match.
Keywords
Keywords (english)
Languagecroatian
URN:NBNurn:nbn:hr:200:848536
Study programmeTitle: Graduate University Study Programme in Computer Engineering
Study programme type: university
Study level: graduate
Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
Type of resourceText
File originBorn digital
Access conditionsOpen access
Terms of useLicence In copyright icon
Created on2019-12-17 08:39:27