Speakers:
Professor Steve Marshall, University of Strathclyde
"Detection of biological cells in noisy images"
Many image processing tasks aim to identify and extract features and patterns in a set of images. Where images are corrupted by noise during the acquisition process, many detection algorithms fail. We propose a method that uses a novel design tool to set parameters for object recognition using generalisations and extensions of mathematical morphology. Given information about the shape and size of objects to be detected, the proposed design tool can be used to set parameters for this algorithm and may be adopted by other researchers such that they can set equivalent parameters for their own routines. The detection algorithm, which is an extension of the Hit-or-Miss Transform (HMT), measures the extent to which a composite structuring element (SE) fits the image at every pixel. A measure of fitness or occupancy for both SEs is calculated in a single pass of the image and objects are marked when the SE satisfies a fitting criteria which is set manually or automatically using our novel design tool. The tool itself can be used to determine a minimum occupancy requirement for any SE such that it can be used to detect features in an image in very noisy conditions. Additionally, this tool allows the detection algorithm to operate in a discriminatory fashion and may also be used to estimate the underlying image noise.
Dr. Pasquale Maffia, Glasgow Biomedical Research Centre
"Real time imaging of immune cells in atherosclerosis and stroke"
T lymphocytes are one of the main cells controlling the immune response and they are present in both atherosclerotic vessels and ischemic brain, but their role in vascular injury and stroke remains to be clarified. In vivo analysis of these lymphocytes in the induction and progression of vascular pathologies is the most instructive way to clarify these processes and will be critical for the development of therapies which might target lymphocytes. We are now focusing to better elucidate multiphoton microscopy potentialities in cardiovascular medicine, combining our experience of tracking lymphocytes in vivo with the use of well established animal models of atherosclerosis and stroke, visualising directly in real time, the cellular interactions underlying development and progression of these pathologies
Dr Paul Herron, Strathclyde Institute of Pharmacy and Biomedical Sciences, University of Strathclyde
"Imaging Polarized Growth of the Mycelial Bacterium, Streptomyces Coelicolor"
Key to understanding the complex life cycle of the spore-forming, filamentous bacterium, Streptomyces coelicolor, is the sequence of events involved in hyphal growth and chromosome segregation. Unlike rod shaped bacteria, growth of Streptomyces is mono-directional where exponential growth is supported through hyphal tip extension and branching. The mechanism by which chromosome replication and segregation is reconciled with mono-directional growth is not known. To address nucleoid colonization of extending hyphal tips, we analyzed the tracking of replication factories, using DnaN-EGFP [1], in S. coelicolor by time-lapse microscopy [2]. Germinating spores displayed a separation of fluorescent foci prior to tip emergence, before the appearance of additional replication factories some distance from the hyphal tip. We have never observed replication DnaN-EGFP at hyphal tips, although use of fluorescence in situ hybridization and nucleoid staining confirmed the existence of chromosomes in tip-proximal hyphal regions. Tip-proximal replication factories followed growing hyphal tips at an equivalent speed, although the rate of movement of more tip-distal replication factories diminished with increasing distance from the hyphal tip. At branch points we observed separation of fluorescent foci into two spots; both the main hypha and branch each received one of the replication factories. However, hyphal branches which displayed growth arrest [1] soon after emergence did not receive replication factories. Taken together these data suggest, that streptomycetes do not replicate their DNA at the hyphal tip although there is, at least an indirect association, between tip growth and DNA replication.
The time consuming nature of manual measurements of polarized growth highlights the need for image processing and feature extraction software that is integrated with a mathematical model that can describe polarized growth and accurately reproduce in silico the growth that we observe in vivo. In this way we can test predictions made by the model with observations of perturbations made to wet experiments. Key to this is the extraction of numerical features from movies of S. coelicolor and their use to construct a mathematical model that acts as a mycelial automaton and can accurately reproduce these movies in silico. These numerical features include tip extension rate, tip growth distance, tip growth angle, branch distance, branch angle movement of FtsZ and DnaN. Mathematical perturbations may be carried out by changing appropriate constants that reflect changes in tip extension and branching brought about by alterations in the content of the anionic phospholipid, cardiolipin.
[1] Ruban-Osmialowska, B., et al., (2007). J. Bacteriol., 188, 7311–7316.
[2] Jyothikumar, J., et al., (2008). Appl. Environ Microbiol., 74, 6774-6781.
Dr Tom Kelsey, University of St. Andrews
"The automatic identification of non-growing follicles in human ovaries"
The human ovary contains a fixed number of non-growing follicles (NGF) established before birth; this number declines with increasing age culminating in the menopause at 50-51 years on average.
NGF populations are estimated using a standard methodology: the ovary is fixed, thin slices (5-20 micrometres) are taken at regular intervals, and these are stained with hematoxylin and eosin (HE). Sample regions are photographed, with the NGFs appearing in these images counted by hand. Assuming an even distribution of NGFs throughout the ovary, the population is estimated by integration. This process is time consuming, and suffers from human mis-classification, integration error due to small sample sizes, and the inconsistent assumption of even distribution. In this talk I present a combined histological and automatic feature detection approach, leading to reduced human and sampling errors at low magnification and which can, in principle, be used to obtain almost exact NGF populations from fully sectioned ovaries.
- - - -
**Finishing with networking opportunities over food and drink **
This event is free of charge.
Please register to help us guage numbers for catering.
*** Programme now available see link below***
