Institutional Repository of Vilnius University
Pilvo aortos tomografijos vaizdų segmentavimo automatizavimas /
Abstract
dc:descriptionAutomation of abdominal aortic segmentation is an important task that can optimize the work of medical staff by many times. Now doctors are losing a lot of work hours marking the aorta in computed tomography images manually. The main goal of this work is to develop a working method that will be able to automatically mark the aorta and its branches in computed tomography images. An analysis of the literature has shown that this task can be solved using either neural networks or traditional image processing methods. The first solution was immediately rejected due to lack of available data. Traditional image processing techniques can detect aortic boundaries, but they have one major drawback in tracking aortic bifurcations (such as at the kidneys or legs) - they mark only one branch and ignore the other. The detail of the abdominal aortic model is very important in assessing the blood flow to this organ and the symptoms of possible diseases. Therefore, the method was supplemented with the DBSCAN clustering algorithm that was used to identify the branches. The decision is made based on two consecutive images of the CT image. After bifurcation detection, each cluster in other CT image images can be analyzed separately. Software was developed for the formulated method to test its practical effectiveness. The proposed solution to the problem has shown good results in the tracking and labeling of the aorta and its branches, but some parts of the algorithms still need to be improved in order to develop a universal solution. The results of the work were presented at the scientific conference " Lithuanian MSc Research in Informatics and ICT", as well as published in a scientific article.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2021
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Palkevič, Vladyslav,
- Contributors dc:contributor
-
- Dzemyda, Gintautas
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD107115249&prefLang=en_US