Unsupervised Discovery of Hidden Geometry in High Dimensional Datasets for Neuronal Arbor Analytics

dc.contributor.advisorRoysam, Badrinath
dc.contributor.committeeMemberHan, Zhu
dc.contributor.committeeMemberPrasad, Saurabh
dc.contributor.committeeMemberMayerich, David
dc.contributor.committeeMemberVaradarajan, Navin
dc.contributor.committeeMemberEriksen, Jason
dc.creatorLu, Yanbin
dc.date.createdMay 2016
dc.date.submittedMay 2016
dc.description.abstractBrain cells such as microglial, neurons, and astrocytes, undergo varieties of morphological and functional changes during their life cycle, or responding to micro environmental perturbation. To better understand the functions of these cells in different status and sub-status, it is essential to study the morphological characteristics of the cells. More importantly, the process of categorizing cells into different status and sub-status serves as a prerequisite for further analysis. In this dissertation, we propose a method for comparative profiling of quantitative arbor morphology data across multiple ensembles of brain cells (neurons, glia) with the overall goal of analyzing alterations/differences in cellular arbors, in a manner that also facilitates qualitative interpretation and model inference. Our method works in three steps. First, we propose a robust unsupervised co-clustering method for the purpose of robustly identifying groups and sub-groups of cells with similar morphologies, and simultaneously identifying the hierarchical grouping patterns among the corresponding arbor measurements. Second, we compare the identified cell groups and morphological measurements using an adaptation of Tibshirani’s Sparse Group LASSO algorithm, allowing us to identify the most significant feature groups, in addition to individual features that underlie the cellular arbor alterations. Finally, for the special common case when batches of images are continuously being acquired under a common imaging protocol, we describe an approach in which the first step can be performed only once, and the resulting models are reused for subsequent imaging experiments, thereby saving computing time. We illustrate our comparative arbor analytics method with data from a neuro-engineering study aimed at profiling glial alterations resulting from the insertion of a neural recording device into the brain tissue. The experiments on microglial and astrocytes perturbation analysis in the Binge Alcohol study also proofed the power of the proposed method in discovering cell types/subtypes, and discovering features of significance in real biological applications.
dc.description.departmentElectrical and Computer Engineering, Department of
dc.format.digitalOriginborn digital
dc.identifier.citationPortions of this document appear in: Kulkarni, P. M., Barton, E., Savelonas, M., Padmanabhan, R., Lu, Y., Trett, K., & Roysam, B. (2015). Quantitative 3-D Analysis of GFAP Labeled Astrocytes from Fluorescence Confocal Images. Journal of Neuroscience Methods, 246, 38-51; and in: Lu, Y., Trett, K., Shain, W., Carin, L., Coifman, R., & Roysam, B. (2013, April). Quantitative profiling of microglia populations using harmonic co-clustering of arbor morphology measurements. In Biomedical Imaging (ISBI), 2013 IEEE 10th International Symposium on (pp. 1360-1363). IEEE; and in: Lu, Y., Carin, L., Coifman, R., Shain, W., & Roysam, B. (2015). Quantitative arbor analytics: unsupervised harmonic co-clustering of populations of brain cell arbors based on L-measure. Neuroinformatics, 13(1), 47-63; and in: Megjhani, M., Rey-Villamizar, N., Merouane, A., Lu, Y., Mukherjee, A., Trett, K., & Roysam, B. (2015). Population-scale three-dimensional reconstruction and quantitative profiling of microglia arbors. Bioinformatics, btv109; and in: Xu, X., Lu, Y., Tung, A. K., & Wang, W. (2006, April). Mining shifting-and-scaling coregulation patterns on gene expression profiles. In Data Engineering, 2006. ICDE'06. Proceedings of the 22nd International Conference on (pp. 89-89). IEEE.
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.subjectArbor Analytics
dc.titleUnsupervised Discovery of Hidden Geometry in High Dimensional Datasets for Neuronal Arbor Analytics
thesis.degree.collegeCullen College of Engineering
thesis.degree.departmentElectrical and Computer Engineering, Department of
thesis.degree.disciplineElectrical Engineering
thesis.degree.grantorUniversity of Houston
thesis.degree.nameDoctor of Philosophy


Original bundle

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

License bundle

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