Abstract
dc:description.abstractMedical imaging studies generate tremendous amounts of data that are reviewedmanually by physicians every day. Medical image segmentation aims to automate theprocess of extracting (segmenting) "interesting" structures from background structuresin the images, saving physicians time and opening the door to more sophisticatedanalysis such as automatically correlating studies over time. This work focuseson segmenting blood vessels (in particular the retinal vasculature), a task that requiresintegrating both local and global properties of the vasculature to produce goodquality segmentations. We use the Tensor Voting framework as it naturally groupsstructures together based on the consensus of locally voting segments. We investigateseveral ways of encoding the image data as tensors and compare our results quantitativelywith a publically available hand-labeled data set. We demonstrate competitiveperformance versus previously published techniques.
Degree
thesis:*- Level thesis:degree_level
- Master's Degree
- Department thesis:degree_department
- Computer Science and Engineering
- Year dc:date.issued
- 2015
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Farmer, Daniel
- Advisors dc:contributor.advisor
-
- Bebis, George
- Nicolescu, Mircea
- Committee member dc:contributor.committeemember
-
- Zaliapin, Ilya
Rights
dc:rights- Statement dc:rights
-
- In Copyright(All Rights Reserved)
- Author(s)
Identifiers
dc:identifier.*- Handle dc:identifier.uri
- http://hdl.handle.net/11714/2528