National and Kapodistrian University of Athens
Μέθοδοι ανάλυσης δισδιάστατων σημάτων και εφαρμογές στην ιατρική
Abstract
dc:descriptionΤhis thesis introduces novel two-dimensional signal analysis methods, focusing on image segmentation, as well as on applications in medicine. The proposed methods utilize textural and intensity information, based on the active contour approach. Textural information is exploited following different approaches, which are associated with alternative balances between computational cost and detail of texture representation: 1) encoding of the spatial distribution of LBPs, which are selected so as to minimize an appropriately defined entropy function, and generation of a transformed image, segmentable by a scalar active contour, 2) embedment of multidimensional feature vectors extracted directly from LBP distributions within a vector-based active contour approach. Τhe increased computational cost associated with vector calculations can be reduced if preceded by initialization based on the scalar approach. Moreover, the novel variable background active contour model (VBAC) is introduced. The formulation of VBAC model ensures that image regions enter selectively into the calculations of model equation terms, so that regions which add to inhomogeneity are excluded. An optimization framework based on genetic algorithms is proposed for the automatic search of active contour parameters. Finally, a classification approach for malignancy risk assessment of thyroid nodules is introduced, based on boundary features extracted from the detected contours. The proposed methods are applied on standard image sets (VisTex, Brodatz, Berkeley), as well as on sets of thyroid ultrasound (US) images for the delineation of thyroid nodules, leading to higher quality segmentations than state-of-the-art segmentation methods. Future perspectives include applications in various domains.
Degree
thesis:*- Grantor dc:publisher
- National and Kapodistrian University of Athens
- Year dc:date
- 2008
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Savelonas, Michail
- Σαβελώνας, Μιχαήλ
Subjects
dc:subject × 14Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/22967