Institutional Repository of Vilnius University
Magnetinio rezonanso tomografijos vaizdų segmentavimas taikant mokymąsi iš keleto pavyzdžių /
Abstract
dc:descriptionFeature-matching generative adversarial networks face several issues: an overly strong discriminator, mode collapse, and the generator's difficulty in creating images with deeper relationships. This work proposes integrating a relation module into the discriminator of feature-matching generative adversarial networks to address these challenges. The relation module provides the generator with more detailed feedback, which should help the generator produce higher-quality images while preventing the discriminator from becoming too strong. The developed feature-matching generative adversarial networks with the relation module were trained to segment brain tissues: cerebrospinal fluid, gray matter, and white matter. Experiments were conducted to evaluate the impact of the relation module on segmentation and image generation. The experiments showed that the model achieved better results in the segmentation task compared to the baseline feature-matching generative adversarial network. The model is capable of segmenting white matter, making it effective for diagnosing stroke, dementia, and cognitive decline. It also exhibits a high sensitivity rate in cerebrospinal fluid segmentation, enabling the detection of changes related to tumors or Alzheimer's disease, but due to the low accuracy in gray matter segmentation, the model has limited applicability in recognizing these conditions.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2025
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Jagučanskytė, Deimantė,
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:ELABAETD239030431&prefLang=en_US