Segmentacija stomatoloških panoramskih rendgenskih snimaka

undergraduate thesis

• Title (english): Segmentation of Stomatologic Panoramic X-ray Images
• Author: Dorian Smoljan
• Mentor: Marko Subašić (mentor)
• Committee member: Marko Subašić (predsjednik povjerenstva)
• Committee member: Denis Milošević (član povjerenstva)
• Committee member: Karlo Koščević (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2021-07-07, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Cilj ovog rada bio je pronaći i testirati odgovarajuću arhitekturu za problem
semantičke segmentacije kada je prisutan manji broj neoznačenih i veći broj označenih
primjera. Predstavljena je cycleGAN arhitektura, te je testirana na konkretnom
problemu semantičke segmentacije ortopanograma. Arhitektura je izabrana
iz razloga što može iskorisiti i neoznačene slike kao pomoć pri treniranju. Dobiveni
rezultati su pokazali da cycleGAN arhitektura uspješno rješava navedeni
problem, no da ipak zaostaje u usporedbi s najnaprednijim diskriminativnim modelima.

Abstract (english)

The goal of this work was to find and test an adequate deep learning architecture
applicable to the problem of semantic segmentation, in the case when we
at our disposal have a limited number of labeled images, but a relatively high
amount of unlabeled images. cycleGAN, a type of generative model was proposed
as solution to the starting problem. It was chosen due to the fact that it can
use unlabeled images as well as labeled to improve training performance. It was
tested on the dataset of stomatologic X-ray images (orthopantomograms), and it
was shown that while it successfully solves the given problem, it still falls behind
most advanced discriminative models.

• Keywords: duboko ucenje; konvolucijske neuronske mreže; semanticka segmentacija; GAN; cycleGAN; ortopanogrami; medicinske snimke
• Keywords (english): deep learning; convolutional neural networks; semantic segmentation; GAN; cycleGAN; orthopantomograms; medical imaging
• Language: croatian
• URN:NBN: urn:nbn:hr:168:943867
• Study programme: Title: Computing; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-14 21:39:19