Binarna klasifikacija histoloških slika korištenjem neuronskih mreža

undergraduate thesis

• Title (english): Binary classification of histological images using neural networks
• Author: Lorena Žužak
• Mentor: Marko Subašić (mentor)
• Committee member: Marko Subašić (predsjednik povjerenstva)
• Committee member: Luka Posilović (član povjerenstva)
• Committee member: Denis Milošević (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2022-07-05, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Cilj rada bila je uspješna klasifikacija histoloških dijagnostičkih slika u dvije vrste upalne bolesti crijeva, ulcerozni kolitis i Crohnovu bolest, koristeći unaprijed trenirane konvolucijske neuronske mreže. Sama implementacija se obavljala u Pythonu, koristeći PyTorch biblioteku. Tijekom rada isproban je pristup podacima s razdvajanjem male količine slikovnih podataka dostupnih na pacijente, a kasnije i pristup kada se osim slikovnih podataka u neuronsku mrežu dovode i numerički klinički podaci. Tijekom rada, korištene su mnoge regularizacijske tehnike, te je isprobano više postavki parametara treniranja na više različitih dimenzija slika. Nakon ove analize, problem razlikovanja između te dvije bolesti na temelju ovih podataka ostaje neriješen, zbog niske točnosti i uspješnosti razlikovanja.

Abstract (english)

This paper aimed to successfully classify irritable bowel disease histological images into either Crohn's disease or ulcerative colitis using pre-trained convolutional neural networks. The implementation itself was carried out in python, using the PyTorch library. During the exploration, the small quantity of available image data was handled according to the patient each image belonged to. Later on, numerical clinical dana was incorporated into the neural network. Many data regularization techniques were used, as well as different training parameters and different image dimensions. After this analysis, the problem of differentiating between the two conditions based on this data remains unsolved, seeing as the accuracy and differentiation success stayed fairly low for the entirety of the exploration.

• Keywords: računalni vid; duboko učenje; upalna bolest crijeva; Crohnova bolest; ulcerozni kolitis; konvolucijske neuronske mreže; unaprijed trenirane neuronske mreže; PyTorch; VGG16
• Keywords (english): computer vision; deep learning; irritable bowel disease; Crohn's disease; ulcerative colitis; convolutional neural networks; pre-trained neural networks; PyTorch; VGG16
• Language: croatian
• URN:NBN: urn:nbn:hr:168:138540
• 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: 2023-01-15 11:29:55