| | SLO | ENG | Cookies and privacy

Bigger font | Smaller font

Show document Help

Title:Detekcija jajčnih mešičkov v ultrazvočnih volumnih z globokimi nevronskimi mrežami na osnovi modela transformer : magistrsko delo
Authors:ID Pečar, Žiga (Author)
ID Potočnik, Božidar (Mentor) More about this mentor... New window
Files:.pdf MAG_Pecar_Ziga_2025.pdf (3,45 MB)
MD5: D8E3101131F77E72EC6E8E3186B0DBA2
 
Language:Slovenian
Work type:Master's thesis/paper
Typology:2.09 - Master's Thesis
Organization:FERI - Faculty of Electrical Engineering and Computer Science
Abstract:V okviru magistrskega dela smo razvili in implementirali tri metode za detekcijo jajčnih mešičkov in jajčnikov v 3D ultrazvočnih volumnih z uporabo globokih nevronskih mrež, osnovanih na arhitekturah tipa transformer. Implementirani sistemi uporabljajo globoko učenje ter arhitekturo transformer in so posebej prilagojeni za obdelavo volumetričnih podatkov. Vse tri rešitve vključujejo pripravo vhodnih ultrazvočnih podatkov, učenje modela na ročno segmentiranih vzorcih ter napovedovanje prisotnosti in oblike mešičkov/jajčnikov v novih podatkih. Za izboljšanje natančnosti smo uporabili različne tehnike bogatenja podatkov ter ločeno obravnavali prisotnost jajčnikov in jajčnih mešičkov. Učinkovitost metod smo ovrednotili s pomočjo kvantitativnih metrik ter vizualno analizo rezultatov nad več volumni iz javno dostopne podatkovne zbirke USOVA3D. Ugotovili smo, da modeli transformer potrebujejo znatno več podatkov, da enačijo ali presežejo zmožnosti konvolucijskih nevronskih mrež.
Keywords:nevronske mreže, 3D segmentacija, globoko učenje, 3D sivinski volumni, model transformer
Place of publishing:Maribor
Place of performance:Maribor
Publisher:[Ž. Pečar]
Year of publishing:2025
Number of pages:1 spletni vir (1 datoteka PDF (IX, 47 str.))
PID:20.500.12556/DKUM-95418 New window
UDC:004.932:004.8.032.26(043.2)
COBISS.SI-ID:255983875 New window
Publication date in DKUM:22.10.2025
Views:118
Downloads:48
Metadata:XML DC-XML DC-RDF
Categories:KTFMB - FERI
:
Copy citation
  
Average score:(0 votes)
Your score:Voting is allowed only for logged in users.
Share:Bookmark and Share



Hover the mouse pointer over a document title to show the abstract or click on the title to get all document metadata.

Licences

License:CC BY-NC-ND 4.0, Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International
Link:http://creativecommons.org/licenses/by-nc-nd/4.0/
Description:The most restrictive Creative Commons license. This only allows people to download and share the work for no commercial gain and for no other purposes.
Licensing start date:17.09.2025

Secondary language

Language:English
Title:Detection of ovarian follicles in ultrasound volumes using transformer-based deep neural networks
Abstract:In this master's thesis, we developed and implemented methods for follicle and ovary detection in 3D ultrasound volumes using deep neural networks based on transformer architectures. The implemented systems leverage deep learning and transformer-based design, specifically tailored for processing volumetric medical data. All three solutions include preprocessing of input ultrasound volumes, training the model on manually segmented samples, and predicting the presence and shape of follicles and ovaries in unseen data. To improve accuracy, various data augmentation techniques were applied, and the presence of ovaries and ovarian follicles were addressed separately. The effectiveness of the methods was assessed through quantitative metrics and visual analysis on multiple volumes from the publicly available USOVA3D database. We found that transformer models require significantly more data to match or surpass the capabilities of convolutional neural networks.
Keywords:neural networks, 3D segmentation, deep learning, 3D grayscale volumes, transformer


Comments

Leave comment

You must log in to leave a comment.

Comments (0)
0 - 0 / 0
 
There are no comments!

Back
Logos of partners University of Maribor University of Ljubljana University of Primorska University of Nova Gorica