Semantička segmentacija novotvorina u slikama magnetske rezonancije

undergraduate thesis

• Title (english): Semantic segmentation of neoplasms in magnetic resonance images
• Author: Lovro Rabuzin
• Mentor: Siniša Šegvić (mentor)
• Committee member: Siniša Šegvić (predsjednik povjerenstva)
• Committee member: Matej Grcić (član povjerenstva)
• Committee member: Ivan Grubišić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2021-07-08, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

U ovom radu primijenjene su metode dubokog učenja u svrhu semantičke segmentacije glioma iz snimki magnetske rezonancije mozga. Napravljen je kratki pregled osnovnih koncepata dubokog učenja i predstavljene su tri arhitekture modela koje sam koristio za segmentaciju: U-Net, SwiftNet i DenseNet121 s ljestvičastim naduzorkovanjem. Modeli su korišteni za segmentaciju tumora u kriškama snimki mozga. Uz to, korištena je i 3D adaptacija modela ResNet18 za segmentaciju u volumenima mozga. Napravljena je usporedba dobivenih rezultata i rezultata iz literature.

Abstract (english)

This paper applies deep learning to the problem of semantic segmentation of glioma in MR brain scans. This paper offers a short overview of basic concepts of deep learning and presents three model architectures that were used: U-Net, SwiftNet and DenseNet121 with ladder-style upsampling. The models were used for semantic segmentation of glioma in individual slices of the brain scan volumes. A 3D adaptation of ResNet18 was used for segmentation in entire brain scan volumes. The paper also compares results obtained using the aforementioned models with results from other papers on the same dataset.

• Keywords: analiza medicinskih snimki; gliom; neuronske mreže; duboko učenje; konvolucijske neuronske mreže; semantička segmentacija; U-Net; SwiftNet; DenseNet; ljestvičasto naduzorkovanje
• Keywords (english): medical image analysis; glioma; neural networks; deep learning; convolutional neural networks; semantic segmentation; U-Net; SwiftNet; DenseNet; ladder-style upsampling
• Language: croatian
• URN:NBN: urn:nbn:hr:168:995339
• 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 21:30:10