Segmentacija panoramskih rendgenskih snimaka ljudske čeljusti

master's thesis

• Title (english): Segmentation of Panoramic X-ray Images of Human Jaw
• Author: Gabriela Dorić
• Mentor: Marko Subašić (mentor)
• Committee member: Marko Subašić (predsjednik povjerenstva)
• Committee member: Sven Lončarić (član povjerenstva)
• Committee member: Krešimir Križanović (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2021-07-12, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Segmentacija slike zadatak je iz područja računalnog vida i svoju primjenu pronalazi u mnogim industrijama poput medicine i industrije autonomnih vozila. U okviru ovog rada istražena je primjena navedene metode u analizi dentalnih slika, odnosno primjena segmentacije na panoramskim rendgenskim snimkama ljudske čeljusti. Objašnjeni su temeljni koncepti dubokog učenja te su opisani korišteni algoritmi detekcije i segmentacija objekata. Za treniranje mreže korištena je state-of-the-art arhitektura Mask RCNN koja paralelno detektira i segmentira objekte na slici. Kao radna okolina za treniranje modela korišten je radni okvir Tensorflow. Model je treniran na umjetno stvorenim snimkama te je evaluiran na stvarnim snimkama. Dobiveni rezultati potvrđuju da je segmentacija objekta koristan alat u dentalnoj industriji te da umjetno generirani podatci mogu nadomjestiti nedostatak označenih podataka te na taj način smanjiti potrošnju vremenskih i novčanih resursa.

Abstract (english)

Image segmentation is a task in the field of digital image processing and computer vision, and as such finds its application in many industries, for example medicine and the autonomous vehicle industry. Within this paper, the application of image segmentation was investigated in the analysis of dental images, i.e. the application of segmentation on panoramic X-rays of the human jaw. The basic concepts of deep learning and used algorithms for detection and segmentation of objects are described. The state-of-the-art Mask RCNN architecture was used to train the network, which detects and segments objects in the image in parallel. The Tensorflow framework was used as the working environment for model training. The model was trained on artificially created images and evaluated on actual images. The obtained results confirm that the image segmentation is a useful tool in the dental industry and that artificially generated data can compensate for the lack of labeled

• Keywords: segmentacija slike; detekcija objekata; duboko učenje; konvolucijske neuronske mreže; Tensorflow
• Keywords (english): image segmentation; object detection; deep learning; CNN; Tensorflow
• Language: croatian
• URN:NBN: urn:nbn:hr:168:898260
• 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 12:07:43