prikaz prve stranice dokumenta Segmentacija epikardijalne masti
  Download
PDF 6.27 MB
master's thesis
Segmentacija epikardijalne masti
2020. urn:nbn:hr:200:592929
Benčević, Marin
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

Benčević, M. (2020). Segmentacija epikardijalne masti (Master's thesis). Retrieved from https://urn.nsk.hr/urn:nbn:hr:200:592929

Benčević, Marin. "Segmentacija epikardijalne masti." Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, 2020. https://urn.nsk.hr/urn:nbn:hr:200:592929

Benčević, Marin. "Segmentacija epikardijalne masti." Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, 2020. https://urn.nsk.hr/urn:nbn:hr:200:592929

Benčević, M. (2020). 'Segmentacija epikardijalne masti', Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, accessed 18 March 2023, https://urn.nsk.hr/urn:nbn:hr:200:592929

Benčević M. Segmentacija epikardijalne masti [Master's thesis]. Osijek: Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek; 2020 [cited 2023 March 18] Available at: https://urn.nsk.hr/urn:nbn:hr:200:592929

M. Benčević, "Segmentacija epikardijalne masti", Master's thesis, Josip Juraj Strossmayer University of Osijek, Faculty of Electrical Engineering, Computer Science and Information Technology Osijek, Osijek, 2020. Available at: https://urn.nsk.hr/urn:nbn:hr:200:592929

Please login to the repository to save this object to your list.
Metadata
TitleSegmentacija epikardijalne masti
Title (english)Epicardial fat segmentation
AuthorMarin Benčević
MentorIrena Galić (mentor)
Committee memberIrena Galić (predsjednik povjerenstva)
Committee memberKrešimir Nenadić (član povjerenstva)
Committee memberHrvoje Leventić (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 country2020-07-23, Croatia
Scientific / art field,
discipline and subdiscipline		TECHNICAL SCIENCES
Computing
Artificial Intelligence
Abstract
Epikardijalna mast (EM) masno je tkivo koje se nalazi izmed¯u srca i perikarda. Volumen i debljina epikardijalnog masnog tkiva povezai su s razliˇcitim bolestima kao što su ateroskleroza, masnoc´a srca, te drugim nepovoljnim kardiovaskularnim dogad¯ajima. EM je nezavisnan indikator rizika od srˇcanih bolesti. Potpuno automatskim i pouzdanim mjerenjem EM iz CT slika mogu se obraditi veliki skupovi CT slika i time napraviti sistematska studija epikardijalne masti. Ovaj rad predlaže potpuno automatsku metodu semantiˇcke segmetnacije epikardijalne masti iz CT slika koriste ´ci duboke neuronske mreže. Koristi se modificirana U-Net arhitektura s ResNet34 koderom. Model je treniran na 20 CT slika srca dobivenih iz dva razlicˇita CT ured¯aja. Rezultat modela su predikcije segmentiranih podruˇcja EM na CT slikama. Kvantitativna evaluacija rezultata pokazuje podudaranje Dice mjerila od 0,70. U radu je pokazano da je mogu´ce posti´ci grubu semantiˇcku segmentaciju EM dubokom neuronskom mrežom ˇcak i uz mali broj podataka uz pomo´c dobre predobrade podatkovnog skupa. Rad pokazuje i ovisnost utjecaja augmentacije podataka na konaˇcni rezultat semantiˇcke segmentacije CT slika.
Abstract (english)
Epicardial fat (EF) is an adipose tissue located between the heard and the pericardium. The volume and thickness of EF is linked with various conditions such as atherosclerosis, visceral adiposity and other cardiovascular events. EF is also an independent cardiovascular disease risk factor. Fully automatic and reliable measurements of EF from CT scans could enable processing large CT image data sets for a systemic EF study. This thesis proposes a method for fully automatic semantic segmentation of EF tissue from CT images using a deep neural network. The model has a modified U-Net architecture with a ResNet34 encoder. The model is trained on 20 CT scans of the heart. The model results in a prediction of segmented EF area on CT images. Quantitive analysis of the model’s results yields a Dice score of 0.70. This thesis shows that it is possible to achieve rough semantic segmentation of EF tissue using a neural network even without large amounts of data, provided the data is preprocessed appropriately. It also shows the impact of data augmentation on segmentation results from CT images.
Keywords
Keywords (english)
Languagecroatian
URN:NBNurn:nbn:hr:200:592929
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 on2020-09-16 10:52:36