Back to search

Institutes outside Greece

Visual analytics for spatially resolved omics data at single cell resolution: methods and applications

Abstract

dc:description

A deeper understanding of an organism’s pathology is essential for the development of treatments. Over centuries of systematic research, clinical researchers have demonstrated that the more information they acquire about cellular properties and their organization within tissue, the better they can understand the functionality of an organism and disease progression. In recent years, the advent of high-resolution imaging techniques has provided researchers with new information at the single-cell level, enabling precise characterization of cells and the study of their spatial distribution in tissue. However, extracting meaningful biological insights from such new and complex data—where experts do not know the intrinsic characteristics of the data nor the patterns they aim to identify—requires an exploratory data analysis approach. The aim of this dissertation is therefore the development of an end-to-end pipeline for the analysis of these high-dimensional cellular images, from raw data preprocessing to the exploration of cellular patterns and their association with clinical characteristics. In Chapter 1, we motivate the importance and necessity of our study, present the main characteristics of the data, discuss preprocessing algorithms, and introduce a visual analytics approach for the exploratory analysis of complex data. Then, we describe a new methodology for preprocessing these high-dimensional cellular images (Chapter 2). In Chapter 2.1, we present the semi-supervised workflow developed for normalizing Imaging Mass Cytometry data across different samples and dimensions. Additionally, we show how binary classification of image pixels into foreground and background can eliminate non-biological variation between tissue samples and improve the traceability of cells with low protein expression levels. In Chapter 2.2, we present a segmentation algorithm aimed at the precise identification of microglia cells, as captured through the Vectra multiplexed immunofluorescence modality. Specifically, a combination of traditional segmentation algorithms (Level-Set, Watershed) and morphological operations is used to create an automated algorithm for accurately identifying microglial cellular boundaries. In Chapters 3 and 4, we propose a visual analytics approach for the analysis of highly complex spatial cellular data, incorporating two data-driven interactive tools, ImaCytE and SpaCeCo. ImaCytE (Chapter 3) enables experts to examine multiple tissue samples and identify new cell types, explore the spatial patterns they form at multiple levels, and stratify specific cell types based on their microenvironmental characteristics. SpaCeCo (Chapter 4) complements ImaCytE and allows experts to determine which cellular features distinguish two cohorts of samples with different clinical characteristics. The comparison workflow is divided into two steps: first, cohorts are compared based on the abundance of different cell types, and second, based on the spatial patterns formed by these cell types. The comparison is performed at both the cohort and sample levels, allowing experts to identify potential outliers within each cohort. In both ImaCytE and SpaCeCo, users can localize each finding within the tissue to verify it and place it in its biological context. In Chapter 5, we demonstrate how parts of our pipeline were applied in a study on Alzheimer’s disease. Specifically, the segmentation algorithm for efficient identification of microglia, the motif glyphs first introduced in ImaCytE for exploring cellular patterns in tissue, and SpaCeCo for comparing Alzheimer’s patients with healthy individuals were utilized. Finally, we developed visual analytics solutions enabling experts to explore in depth spatial cellular data and generate hypotheses about the origins of tissue functionality in diseased and healthy states. Beyond core exploratory tasks, experts are also enabled to preprocess their data, creating an end-to-end pipeline for the analysis of multiplexed tissue images at cellular resolution. Overall, the work presented in this dissertation lays the foundation for a visual analytics approach to high-dimensional imaging analysis, enabling detailed study of cells and tissues.

Degree

thesis:*
Grantor dc:publisher
Institutes outside Greece
Year dc:date
2022

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Somarakis, Antonios
  • Σωμαράκης, Αντώνιος

Subjects

dc:subject × 32

Rights

Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/61058

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Somarakis, Antonios; Σωμαράκης, Αντώνιος. Visual analytics for spatially resolved omics data at single cell resolution: methods and applications. Institutes outside Greece, 2022. http://hdl.handle.net/10442/hedi/61058