Author: Sami Almutairi
Almutairi, Sami, 2017 Evaluation of Statistical Region Merging Segmentation for heart and spleen contour delineation in CT images, Flinders University, School of Computer Science, Engineering and Mathematics
Terms of Use: This electronic version is (or will be) made publicly available by Flinders University in accordance with its open access policy for student theses. Copyright in this thesis remains with the author. You may use this material for uses permitted under the Copyright Act 1968. If you are the owner of any included third party copyright material and/or you believe that any material has been made available without permission of the copyright owner please contact copyright@flinders.edu.au with the details.
Segmentation is the key step towards extracting information from medical images for further
analysis. This study investigates the accuracy of organ segmentation from CT images using
the Statistical Region Merging (SRM) segmentation technique. The experiments were
conducted using data set of 12 consecutive CT slices for heart and 8 CT slices for spleen. The
data sets were previously manually segmented by an expert in human anatomy and these sets
constituted the ground truth to judge the accuracy of an automatic segmentation using SRM.
The results of SRM segmentation technique depend on a parameter (Q) which value
determines granularity of the outcome. A wide range of Q values was investigated in relation
to the accuracy of segmented heart and spleen. The accuracy was measured using the Dice
and Jaccard indexes. The results show that the SRM segmentation technique can potentially
provide high accuracy segmentation of 80% and above for both heart and spleen. However,
neither heart nor spleen can be segmented at this level of precision as a single component
and an additional post-processing is needed to merge eligible regions into one piece
corresponding to the whole organ.
Keywords: Statistical Region Merging Segmentation
Subject: Engineering thesis
Thesis type: Masters
Completed: 2017
School: School of Computer Science, Engineering and Mathematics
Supervisor: Dr. Mariusz Bajger