3D segmentacija CT snimaka bubrežnog kamenca

master's thesis

• Title (english): 3D segmentation of kidney stone CT images
• Author: Matea Đonlić
• Mentor: Damir Seršić (mentor)
• Committee member: Damir Seršić (predsjednik povjerenstva)
• Committee member: Mladen Vučić (član povjerenstva)
• Committee member: Marko Subašić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2014-07-09, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Ovaj diplomski rad usmjeren je na implementaciju i usporedbu različitih metoda 3D segmentacije s primjenom na problem segmentacije CT snimaka bubrežnog kamenca.
Budući da se sve metode 3D segmentacije provode nad diskretnom trodimenzionalnom rešetkom, ulazni podaci (skup datoteka u DICOM formatu) obrađeni su i oblikovani tako da je omogućen odabir područja interesa i iz njega stvaranje diskretne rešetke.
U sklopu rada, implementirane su metode amplitudne segmentacije, segmentacije izrastanjem područja te segmentacije temeljene na lokalnoj faznoj koherenciji. Kao ocjena uspješnosti segmentacije korištena je mjera sličnosti između napravljenog modela i referentnog modela izvađenog kamenca. Mjera sličnosti dobivena je provođenjem postupka registracije modela korištenjem ICP algoritma. Također, napravljena je usporedba s rezultatima segmentacija provedenih korištenjem dostupnih programskih alata za 3D medicinsku segmentaciju. Napravljena je i usporedba brzine izvođenja implementiranih segmentacija te ocjena uspješnosti segmentacija u ovisnosti o subjektivnim odabirima područja interesa.

Abstract (english)

This thesis focuses on the implementation and comparison of different 3D segmentation methods along with their application to the medical problem of kidney stone CT image segmentation.
Considering the fact that the all segmentation methods require a discrete three-dimensional grid, the input data in DICOM format had to be processed and shaped in the form suitable for the process of region of interest (ROI) selection and the creation of the three-dimensional grid.
Three 3D segmentation methods were presented and implemented in this thesis -- an amplitude segmentation method, a region growing segmentation and a segmentation method based on the local phase coherence. For the evaluation of implemented segmentation methods, created models were compared with surgically removed kidney stone and also with the models created using few of the available software tools for 3D medical segmentation. Models were compared based on the similarity measure obtained via ICP algorithm. Also, an execution time analysis for different segmentation methods was made as well as the comparison of the segmentation performance depending on the subjective choices of ROI.

• Keywords: automatska 3D segmentacija; amplitudna segmentacija; izrastanje područja; lokalna fazna koherencija; ICP algoritam
• Keywords (english): automatic 3D segmentation; amplitude thresholding; region growing; local phase coherence; ICP algorithm
• Language: croatian
• URN:NBN: urn:nbn:hr:168:144249
• Study programme: Title: Information and Communication Technology; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka informacijske i komunikacijske tehnologije (magistar/magistra inženjer/inženjerka informacijske i komunikacijske tehnologije)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2018-02-06 08:38:46