prikaz prve stranice dokumenta Segmentacija koljena difuzijskim modelima iz 3D slika magnetske rezonance
  Download
PDF 5.07 MB
master's thesis
Segmentacija koljena difuzijskim modelima iz 3D slika magnetske rezonance
Osijek: Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, 2024. urn:nbn:hr:200:562980
Perić, Juraj
Josip Juraj Strossmayer University of Osijek
Faculty of Electrical Engineering, Computer Science and Information Technology Osijek
Department of Software Engineering
Chair of Visual Computing

Institutional repository: Faculty of Electrical Engineering, Computer Science and Information Technology Osijek

Cite this document

Perić, J. (2024). Segmentacija koljena difuzijskim modelima iz 3D slika magnetske rezonance (Master's thesis). Osijek: Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek. Retrieved from https://urn.nsk.hr/urn:nbn:hr:200:562980

Perić, Juraj. "Segmentacija koljena difuzijskim modelima iz 3D slika magnetske rezonance." Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, 2024. https://urn.nsk.hr/urn:nbn:hr:200:562980

Perić, Juraj. "Segmentacija koljena difuzijskim modelima iz 3D slika magnetske rezonance." Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, 2024. https://urn.nsk.hr/urn:nbn:hr:200:562980

Perić, J. (2024). 'Segmentacija koljena difuzijskim modelima iz 3D slika magnetske rezonance', Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, accessed 03 October 2024, https://urn.nsk.hr/urn:nbn:hr:200:562980

Perić J. Segmentacija koljena difuzijskim modelima iz 3D slika magnetske rezonance [Master's thesis]. Osijek: Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek; 2024 [cited 2024 October 03] Available at: https://urn.nsk.hr/urn:nbn:hr:200:562980

J. Perić, "Segmentacija koljena difuzijskim modelima iz 3D slika magnetske rezonance", Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, Osijek, 2024. Available at: https://urn.nsk.hr/urn:nbn:hr:200:562980

Metadata
TitleSegmentacija koljena difuzijskim modelima iz 3D slika magnetske rezonance
Title (english)Knee segmentation using diffusion models from 3D magnetic resonance images
AuthorJuraj Perić
MentorIrena Galić (mentor)
MentorMarija Habijan (sumentor)
Committee memberIrena Galić (predsjednik povjerenstva)
Committee memberHrvoje Leventić (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 country2024-09-13, Croatia
Scientific / art field,
discipline and subdiscipline		TECHNICAL SCIENCES
Computing
Artificial Intelligence
Abstract
Segmentacija koljena iz slika magnetne rezonance važna je za praćenje progresije bolesti poput osteoartritisa i planiranje odgovarajuće terapije. U ovom radu predstavljen je model za automatsku segmentaciju koljena korištenjem difuzijskih modela, metode koja je u posljednje vrijeme postala popularna u području generiranja slika. Model koristi U-Net arhitekturu s modifikacijama prilagođenim za difuzijske procese, uključujući dinamičko uvjetno kodiranje i FF-pretvarač koji su predloženi u radu MedSegDiff: Medical Image Segmentation with Diffusion Probabilistic Model. Trening modela proveden je na skupu podataka Osteoarthritis Initiative (OAI), koji sadrži 3D MRI slike koljena s pripadajućim oznakama kostiju i hrskavica. Model je evaluiran korištenjem metrika F1-Score i Intersection over Union (IoU), pri čemu su postignuti rezultati od 89,21 % i 82,1 % nakon naknadne primjene Gaussovog zamućenja. Ovi rezultati su usporedivi s postojećim metodama segmentacije koljena. Iako model pokazuje visoku razinu preciznosti i otvorenost za moguću nadogradnju, njegova primjena u praksi može biti ograničena zbog dugog vremena segmentacije.
Abstract (english)
Knee segmentation from magnetic resonance imaging is crucial for monitoring the progression of diseases like osteoarthritis and planning appropriate therapy. This paper presents a model for automatic knee segmentation using diffusion models, a method that has recently gained popularity in the field of image generation. The model employs a U-Net architecture with modifications tailored for diffusion processes, including dynamic conditional encoding and the FF-Parser proposed in the paper MedSegDiff: Medical Image Segmentation with Diffusion Probabilistic Model. The model was trained on the Osteoarthritis Initiative (OAI) dataset, which contains 3D MRI knee images with corresponding annotations of bones and cartilage. The model was evaluated using F1-Score and Intersection over Union (IoU) metrics, achieving results of 89.21% and 82.1% after applying post-processing with Gaussian blurring. These results are comparable with existing knee segmentation methods. While the model demonstrates a high level of accuracy and potential for further improvement, its practical application may be limited due to the long segmentation time.
Keywords
Keywords (english)
Languagecroatian
URN:NBNurn:nbn:hr:200:562980
Study programmeTitle: Graduate University Study Programme in Computer Engineering
Study programme type: university
Study level: graduate
Academic / professional title: sveučilišni/a magistar/magistra inženjer/inženjerka računarstva (sveučilišni/a magistar/magistra inženjer/inženjerka računarstva)
Type of resourceText
File originBorn digital
Access conditionsOpen access
Terms of useLicence In copyright icon
Created on2024-09-19 07:52:32