Shape Priors for Segmentation of Mouse Brain in Gene Expression Images

dc.contributor.advisorKakadiaris, Ioannis A.
dc.contributor.committeeMemberCarson, James P.
dc.contributor.committeeMemberDeng, Zhigang
dc.contributor.committeeMemberEick, Christoph F.
dc.contributor.committeeMemberJu, Tao
dc.contributor.committeeMemberShah, Shishir Kirit
dc.creatorLe, Yen H. 1984- 2014 2014
dc.description.abstractThe quantification and comparison of gene expression data across images plays a key role in understanding the functional network of various genes. To enable studying of relationships between a large number of images, automated methods that can segment gene expression images into distinct anatomical regions/sub-regions are needed. Automated segmentation of mouse brain gene expression images is a challenging problem mainly due to the complexity of gene expression appearance: (i) the lack of visible edge cues for the anatomical regions, (ii) the inhomogeneous of intensity pattern inside each anatomical region, and (iii) the variation of intensity pattern of the same region across images. Therefore, the use of geometric priors and appearance cues can potentially help in accurate segmentation of gene expression images. The goal of this thesis is to develop segmentation methods that incorporate shape priors and appearance cues and apply them to gene expression image data. The specific objectives are: (i) to incorporate statistical shape information to segment gene expression images, (ii) to learn salient model points that will be selected to provide appearance cues for segmentation methods to handle images that complex appearance (e.g., gene expression images), (iii) to improve the representation ability of statistical shape models to represent a larger range of shapes, and (iv) to evaluate the proposed methods on the segmentation of gene expression images. Corresponding to each of the first three objectives, three methods have been proposed. They all outperform the state-of-the-art on a challenging problem of segmenting gene expression images of mouse brain. The best performance is achieved by PDM-ENLOR (Point Distribution Model-based ENsemble of Local Regressors): the overall mean and standard deviation (over all 14 anatomical regions and all test images) of Dice coefficient overlap were 88.1 % +/- 9.5 % and of Hausdorff distance were 0.235 mm +/- 0.100 mm. This dissertation contributes a method for determining a set of salient points for using in a given segmentation algorithm and a method that increases the flexibility of the statistical shape model and handles the detection errors simultaneously. These data-driven methods are generic and can be applied to other similar problems.
dc.description.departmentComputer Science, Department of
dc.format.digitalOriginborn digital
dc.identifier.citationPortions of this document have appeared in: Le Y.H., Kurkure U., Kakadiaris I.A. (2013) PDM-ENLOR: Learning ensemble of local PDM-based regressions. In Proc. IEEE Conference on Computer Vision and Pattern Recognition, pp 1878-1885. And in: Le Y.H., Kurkure U., Paragios N., Ju T., Carson J.P., Kakadiaris I.A. (2012) Similarity-Based Appearance-Prior for Fitting a Subdivision Mesh in Gene Expression Images. In: Ayache N., Delingette H., Golland P., Mori K. (eds) Medical Image Computing and Computer-Assisted Intervention -MICCAI 2012. MICCAI 2012. Lecture Notes in Computer Science, vol 7510. Springer, Berlin, Heidelberg. DOI: 10.1007/978-3-642-33415-3_71
dc.rightsThe author of this work is the copyright owner. UH Libraries and the Texas Digital Library have their permission to store and provide access to this work. UH Libraries has secured permission to reproduce any and all previously published materials contained in the work. Further transmission, reproduction, or presentation of this work is prohibited except with permission of the author(s).
dc.subjectStatistical shape model
dc.subjectGene expression images
dc.titleShape Priors for Segmentation of Mouse Brain in Gene Expression Images
dc.type.genreThesis of Natural Sciences and Mathematics Science, Department of Science of Houston of Philosophy


Original bundle

Now showing 1 - 1 of 1
Thumbnail Image
24.04 MB
Adobe Portable Document Format

License bundle

Now showing 1 - 1 of 1
No Thumbnail Available
1.84 KB
Plain Text