Detekcija oštećenog tkiva na CT slikama pluća pomoću umjetne neuronske mreže

undergraduate thesis

• Title (english): Detection of damaged tissue on CT lung images using artificial neural networks
• Author: David Ivušić
• Mentor: Sonja Grgić (mentor)
• Committee member: Sonja Grgić (predsjednik povjerenstva)
• Committee member: Marko Bosiljevac (član povjerenstva)
• Committee member: Jelena Božek (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 ovog rada je prikazati kako umjetne neuronske mreže mogu pomoći zdravstvenim djelatnicima pri dijagnozi bolesti te tako olakšati i ubrzati proces liječenja pacijenta. Za taj zadatak korištena je darknet implementacija YOLO (You Only Look Once) arhitekture čija je glavna prednost nad ostalim mrežama za detekciju objekata njegova brzina. Osim YOLO arhitekture opisane su i R-CNN (Regions with CNN features) arhitekture. Model istreniran u ovom radu ima mAP (mean average precision) 59.14%, a testiranje na 100 slika pokazuje da model dosta dobro detektira česta oštećenja, a lošije oštećenja koja se rijetko pojavljuju ili koja se mogu zamijeniti s normalnim nalazom.

Abstract (english)

The aim of this paper is to show how artificial neural networks can help health professionals in diagnosing the disease and thus facilitate and speed up the process of treating the patient. For this task, the darknet implementation of the YOLO (You Only Look Once) architecture is used, whose main advantage over other networks for object detection is its speed. In addition to the YOLO architecture, R-CNN (Regions with CNN features) architectures are also described. The model trained in this paper has a mAP (mean average precision) of 59.14% and testing on 100 images shows that the model detects frequent damages quite well, however damages that rarely occur or that can be mistaken for a normal finding are not detected as well.

• Keywords: umjetna inteligencija; umjetni neuron; prijenosne funkcije; višeslojna umjetna neuronska mreža; propagacija unatrag; konvolucijske neuronske mreže; konvolucija; R-CNN; Fast R-CNN; Faster R-CNN; YOLO; darknet; kompjutorizirana tomografija; pluća
• Keywords (english): artificial intelligence; artificial neuron; activation functions; multi-layer neural network; backpropagation; convolutional neural network; convolution; R-CNN; Fast R-CNN; Faster R-CNN; YOLO; darknet; computed tomography; lungs
• Language: croatian
• URN:NBN: urn:nbn:hr:168:554724
• 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 10:22:55