Rekonstrukcija iz projekcija hibridnim pristupom

master's thesis

• Title (english): Reconstruction from Projections Using Hybrid Approach
• Author: Marija Vujica
• Mentor: Damir Seršić (mentor)
• Committee member: Damir Seršić (predsjednik povjerenstva)
• Committee member: Ana Sović Kržić (član povjerenstva)
• Committee member: Marko Subašić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2021-09-16, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

PET je uobičajena medicinska tehnika snimanja koja detektira aktivnosti tijela na molekularnoj razini.
Zadatak programskih potpora, kao i glavni cilj ovog rada jest rekonstrukcija niza slika snimanog tijela koristeći prikupljene projekcije i znanje o prirodi cjelokupnog sustava snimanja.
U radu je najprije opisan postupak prikupljanja projekcija objekta snimanog PET uređajem. Zatim su navedene konvencionalne metode rekonstukcije poput projekcije unatrag, iterativnih metoda i drugih.
Zbog potrebe za poboljšanjima postojećih sustava medicinskih snimanja, u kontekstu rekonstrukcije PET snimaka nastoje se koristiti prednosti i kvalitete metoda strojnog učenja. Stoga su nadalje navedeni standardni pristupi strojnog učenja u konkretnoj problematici.
U konačnici je predložena i isprobana metoda dubokog učenja koja nudi razna poboljšanja rekonstrukcija, nakon čega slijedi demonstracija rezultata dobivenih na temelju realnih snimaka.

Abstract (english)

PET is a common medical acquisition technique that detects body activity on a molecular level.
Role of computer software in this area is being main task of this academic research, which is to reconstruct series of images of captured object using acquired projection data and overall scanning system knowledge.
Foremost, in this assignment there is a description of projection acquisition process given by PET scanning machine. Afterwards, some conventional reconstruction methods are listed, such as backprojection, iterative methods and so on.
Due to need of having existent medical scanning systems consecutively improved, there is an idea of embedding qualities and advantage of machine learning principles in PET reconstruction problems. Hence, there were further listed some standard machine learning approaches in this specific issue.
At last, a deep learning method that in some ways improves reconstructions was proposed, tested and the results were demonstrated on a set of practical real-world dataset.

• Keywords: Pozitronska emisijska tomografija; rekonstrukcija; Strojno učenje
• Keywords (english): Positron emission tomography; reconstruction; Machine learning
• Language: croatian
• URN:NBN: urn:nbn:hr:168:394758
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-13 16:30:28