Methods for analysing endothelial cell shape and behaviour in relation to the focal nature of atherosclerosis

• Main Author: Iftikhar, Saadia
• Other Authors: Bharath, Anil ; Cashman, Peter
• Published: Imperial College London 2011
• Subjects: 660.6
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.537191

The aim of this thesis is to develop automated methods for the analysis of the spatial patterns, and the functional behaviour of endothelial cells, viewed under microscopy, with applications to the understanding of atherosclerosis. Initially, a radial search approach to segmentation was attempted in order to trace the cell and nuclei boundaries using a maximum likelihood algorithm; it was found inadequate to detect the weak cell boundaries present in the available data. A parametric cell shape model was then introduced to fit an equivalent ellipse to the cell boundary by matching phase-invariant orientation fields of the image and a candidate cell shape. This approach succeeded on good quality images, but failed on images with weak cell boundaries. Finally, a support vector machines based method, relying on a rich set of visual features, and a small but high quality training dataset, was found to work well on large numbers of cells even in the presence of strong intensity variations and imaging noise. Using the segmentation results, several standard shear-stress dependent parameters of cell morphology were studied, and evidence for similar behaviour in some cell shape parameters was obtained in in-vivo cells and their nuclei. Nuclear and cell orientations around immature and mature aortas were broadly similar, suggesting that the pattern of flow direction near the wall stayed approximately constant with age. The relation was less strong for the cell and nuclear length-to-width ratios. Two novel shape analysis approaches were attempted to find other properties of cell shape which could be used to annotate or characterise patterns, since a wide variability in cell and nuclear shapes was observed which did not appear to fit the standard parameterisations. Although no firm conclusions can yet be drawn, the work lays the foundation for future studies of cell morphology. To draw inferences about patterns in the functional response of cells to flow, which may play a role in the progression of disease, single-cell analysis was performed using calcium sensitive florescence probes. Calcium transient rates were found to change with flow, but more importantly, local patterns of synchronisation in multi-cellular groups were discernable and appear to change with flow. The patterns suggest a new functional mechanism in flow-mediation of cell-cell calcium signalling.