Author(s):
Martins, Leonardo Pedro Donas-Boto de Vilhena
Date: 2020
Persistent ID: http://hdl.handle.net/10362/107062
Origin: Repositório Institucional da UNL
Project/scholarship:
info:eu-repo/grantAgreement/FCT/SFRH/SFRH%2FBD%2F88987%2F2012/PT;
info:eu-repo/grantAgreement/FCT/3599-PPCDT/PTDC%2FBBB-MET%2F1084%2F2012/PT;
Subject(s): Image Processing; Image Simulation; Cell Segmentation; Cell Tracking; Domínio/Área Científica::Engenharia e Tecnologia::Engenharia Eletrotécnica, Eletrónica e Informática
Description
Recent advances in microscopy imaging technology have allowed the characterization of the dynamics of cellular processes at the single-cell and single-molecule level. Particularly in bacterial cell studies, and using the E. coli as a case study, these techniques have been used to detect and track internal cell structures such as the Nucleoid and the Cell Wall and fluorescently tagged molecular aggregates such as FtsZ proteins, Min system proteins, inclusion bodies and all the different types of RNA molecules. These studies have been performed with using multi-modal, multi-process, time-lapse microscopy, producing both morphological and functional images. To facilitate the finding of relationships between cellular processes, from small-scale, such as gene expression, to large-scale, such as cell division, an image processing toolbox was implemented with several automatic and/or manual features such as, cell segmentation and tracking, intra-modal and intra-modal image registration, as well as the detection, counting and characterization of several cellular components. Two segmentation algorithms of cellular component were implemented, the first one based on the Gaussian Distribution and the second based on Thresholding and morphological structuring functions. These algorithms were used to perform the segmentation of Nucleoids and to identify the different stages of FtsZ Ring formation (allied with the use of machine learning algorithms), which allowed to understand how the temperature influences the physical properties of the Nucleoid and correlated those properties with the exclusion of protein aggregates from the center of the cell. Another study used the segmentation algorithms to study how the temperature affects the formation of the FtsZ Ring. The validation of the developed image processing methods and techniques has been based on benchmark databases manually produced and curated by experts. When dealing with thousands of cells and hundreds of images, these manually generated datasets can become the biggest cost in a research project. To expedite these studies in terms of time and lower the cost of the manual labour, an image simulation was implemented to generate realistic artificial images. The proposed image simulation toolbox can generate biologically inspired objects that mimic the spatial and temporal organization of bacterial cells and their processes, such as cell growth and division and cell motility, and cell morphology (shape, size and cluster organization). The image simulation toolbox was shown to be useful in the validation of three cell tracking algorithms: Simple Nearest-Neighbour, Nearest-Neighbour with Morphology and DBSCAN cluster identification algorithm. It was shown that the Simple Nearest-Neighbour still performed with great reliability when simulating objects with small velocities, while the other algorithms performed better for higher velocities and when there were larger clusters present.
