Přístupnostní navigace
E-application
Search Search Close
Bachelor's Thesis
Author of thesis: Matúš Smolka
Acad. year: 2025/2026
Supervisor: Ing. Radim Kořínek, Ph.D.
Reviewer: Ing. Jiří Kratochvíla, Ph.D.
This thesis presents a multiparametric pipeline for liver segmentation from preclinical magnetic resonance imaging (MRI) data acquired from mice without the use of contrast agents. Multiple quantitative MRI-derived parameters are combined into voxel-wise feature vectors, enabling tissue characterization in multiparametric feature space. Segmentation is performed using an unsupervised Gaussian Mixture Model followed by morphology-driven post-processing to improve the anatomical consistency of the resulting liver masks. The developed framework further incorporates a semi-automated modular processing pipeline, interactive graphical tools for cluster selection, and dedicated three-dimensional visualization tools supporting qualitative evaluation of segmentation performance. The proposed method was evaluated on a cohort of preclinical mouse MRI scans using standard overlap and boundary-based segmentation metrics. The results demonstrate stable segmentation performance in central liver regions, while remaining inaccuracies are observed primarily at organ boundaries and in thin peripheral structures. Overall, the results indicate that multiparametric unsupervised segmentation represents a feasible approach for liver extraction in preclinical MRI under constrained data conditions.
Magnetic Resonance Imaging (MRI), Multiparametric MRI, Liver Segmentation, Preclinical Imaging, Image Segmentation, Gaussian Mixture Model, Unsupervised Clustering, Quantitative MRI, Relaxometry, Medical Image Processing
Date of defence
17.06.2026
Result of the defence
Defended (thesis was successfully defended)
Grading
C
Process of defence
Student prezentoval výsledky své práce a komise byla seznámena s posudky. Doc. Kudrna se doptal k druhé otázce od oponenta: Proč jste to nevyzkoušel? Ing. Vitouš se doptal, zda se student díval na to, jak jsou feature mapy na sobě závislé. Ing. Ředina se zeptal, zda byl počítán DICE per voxel a zda by nešlo udělat nějaké srovnání s jinými existujícími segmentačními nástroji. Student obhájil bakalářskou práci a odpověděl na otázky členů komise a oponenta.
Language of thesis
English
Faculty
Fakulta elektrotechniky a komunikačních technologií
Department
Department of Biomedical Engineering
Study programme
Biomedical Technology and Bioinformatics (BPC-BTB)
Composition of Committee
doc. Ing. Petr Kudrna, Ph.D. (předseda) Ing. Markéta Jakubíčková, Ph.D. (místopředseda) MUDr.Ing. Richard Ředina (člen) Ing. Martin Králík (člen) Ing. Jiří Vitouš (člen) doc. Ing. Radim Kolář, Ph.D. (člen)
Supervisor’s reportIng. Radim Kořínek, Ph.D.
Grade proposed by supervisor: B
Reviewer’s reportIng. Jiří Kratochvíla, Ph.D.
Grade proposed by reviewer: B
Responsibility: Mgr. et Mgr. Hana Odstrčilová