This research project is concerned with automated analysis of microscopic images used in clinical pathology for diagnosing disease. Application of computer vision methods can improve the accuracy, reliability and availability of tests, reduce the associated costs and ultimately improve patient outcomes. Three different areas of pathology are covered: 1. identification of clustered nuclei and detection of chromosomal abnormalities in DAPI-stained samples, 2. diagnosis of auto-immune diseases from indirect immuno fluorescence (IIF) images, and 3. detection of dividing nuclei in H&E stained histopathology sections. Despite the diversity of these application domains, the techniques used for their analysis are similar. For cluster identification in DAPI images we focus on object shape and extend existing methods of shape analysis with novel measurements of the boundary profile which detect notches between overlapping nuclei in a cluster. For abnormality detection we focus on texture and develop a novel decision-tree dictionary for patch quantisation. We continue to focus on texture for IIF images, developing suitable isotropic measurements as well as exploring the connections between classification of individual cells and whole patient samples. Detection of dividing cells in tissue sections requires a combined assessment of shape, texture and colour in order to fully represent all relevant facets of the object. Here we develop a method for stain normalisation which efficiently compensates for batch variations in stain strength and proportions, followed by a full pipe-line of segmentation, feature extraction and classification, resolving issues of class imbalance implicit in detection of rare objects. We develop an efficient and effective segmentation method, which is free of weight parameters and adaptable for use in different imaging modalities. We explore a variety of classifer types and ensemble structures, and suggest promising directions of future development in the broad application area of pathology image analysis.

Positron emission tomography (PET) enables noninvasive tumour imaging, as changes in metabolic activity secondary to therapy can be measured before changes in tumour size are evident on standard anatomic imaging. Two imaging approaches representing proliferation dependent and independent technologies are evolving as potential methods for assessing growth signalling and, thus, treatment response: [18F]3'-deoxy-3'-fluorothymidine (FLT) and [11C]choline. The validity of the former in patients with pancreatic cancer is unproven and likewise, the role of the latter in response to androgen deprivation/radiotherapy in prostate cancer (PCa) remains unexplored. Using a variety of approaches, the aim of this thesis was to provide an understanding of the role of these tracers in lesion detection and response assessment in patients by PET/computed tomography (PET/CT). Given the high physiological hepatic localisation of FLT, a recently reported kinetic spatial filtering (KSF) algorithm was evaluated as a way to de-noise abdominal FLT-PET data from patients with advanced pancreatic cancer. Application of KSF led to improved lesion detection. FLT uptake (SUV60,max) significantly increased in mid-treatment (gemcitabine based) progressors (p=0.04). In this limited number of patients, reduction in FLT uptake did not predict overall survival. The role of [11C]choline PET/CT in lesion detection and response in prostate cancer (PCa) was also investigated using semi-quantitative and quantitative methods. As a prelude to the quantitative imaging studies, it was established that irreversible tracer uptake characterised tumour (breast cancer) [11C]choline kinetics. Similar irreversible uptake characterised PCa. An important finding was that tumour [11C]choline uptake (in 29 PCa patients) correlated with choline kinase (CHK) expression but not proliferation, as assessed by Ki67 labelling index. Immunohistochemistry of the above patients' prostate cores with CHKα antibody demonstrated a spectrum of CHKα expression, ranging from expression in prostatic-intraepithelial-neoplasia to low to high expression in malignant cores. These findings were further corroborated in a larger cohort of 75 malignant cores derived from non-imaging studies. Having established [11C]choline as a proliferation independent marker of growth, its role in assessing treatment response was investigated. [11C]choline PET was sensitive to metabolic changes within prostate tumours following androgen deprivation and radical radiotherapy. While promising data were obtained with [11C]choline PET, the radiotracer is subject to metabolic degradation complicating data analysis. To this end, a novel metabolically stable analogue of choline ([18F]fluoromethyl-[1,2-2H4]-choline ([18F]D4FCH)) was transitioned into volunteers and patients to study its pharmacokinetics and preliminary diagnostic potential. This tracer embodies deuterium isotope substitution as a means to discourage systemic metabolism. The radiotracer had favourable dosimetry (effective-dose: 0.025mSv/MBq) and safety. Preliminary results in non-small cell lung cancer showed that the tracer is taken up in tumours. Further studies are warranted to characterise this new tracer in different tumour types. As a prelude to imaging cancer cell death in tumours, a caspase-3 specific radiotracer, [18F](S)-1-((1-(2-fluoroethyl)-1H-[1,2,3]-triazol-4-yl)methyl)-5-(2(2,4- difluorophenoxymethyl)-pyrrolidine-1-sulfonyl) isatin ([18F]ICMT-11) was also transitioned into volunteers. The radiotracer had favourable dosimetry (effective-dose: 0.025mSv/MBq) and safety. In summary, FLT-PET/CT combined with KSF and [11C]choline PET/CT were shown to be promising methods for imaging early treatment response in patients. Further work will be required to evaluate the clinical relevance of these data in terms of overall patient outcome. Furthermore, a new choline-based radiotracer and a caspase-3 specific radiotracer have been transitioned into humans.

Respiratory motion degrades PET spatial resolution and image quality limiting the potential benefits from increased resolution. Motion correction is complicated by limitations of the poor statistical quality of PET data and there is still not a robust method available. Motion correction can be implemented at different stages of data processing either during or after reconstruction and may result in substantial improvements in image quality. The recent development of whole body PET-MRI scanners might provide a potential solution for motion correction since internal organ motion could be measured concurrently with PET using MRI. However, although there have been various proposed methods for motion correction, there is not sufficient evidence in the current literature to answer which method is better in clinical practice and investigating the impact of motion correction on lesion detectability. The aim of this thesis is to assess respiratory motion correction in terms of its quantitative accuracy and detectability performance to determine its potential for improved and early cancer diagnosis. This thesis is based on numerical 4D simulated PET data using real MRI data. Motion correction is investigated based on MRI-derived motion fields as could be obtained from a simultaneous PET-MRI acquisition. As a first step, this thesis aims to understand the behaviour of different approaches to motion-corrected image reconstruction in terms of convergence rate and the properties of the reconstructed images obtained. This thesis then deals with the impact of respiratory motion on lesion detectability. A comprehensive assessment is performed using different amplitudes for lesion displacement due to respiration and different respiration patterns derived from actual patient respiratory traces. The impact on the detectability is compared with that achievable by a higher resolution scanner in order to investigate the importance of correcting for motion to realise the benefit from the increased resolution of future PET scanners.

The major functional units of the lung called alveoli, are located in the periphery of the lung. Despite their functional importance, it has been difficult to evaluate their structure and development. Until the advent of 3-Helium magnetic resonance (3HeMR), it was not possible to directly assess peripheral lung structure in a living individual. When this study commenced, the prevailing hypothesis was that human alveolarization was complete by 3 years. It was believed that preterm birth would lead to persisting alveolar damage. The role of other factors affecting human lung development were not clearly understood. In this work, I describe the use of 3HeMR to : 1. Examine the current hypothesis regarding normal alveolar development, 2. Determine whether birth at very preterm gestation leads to long-term alveolar damage and 3. Evaluate factors affecting human alveolar development. First, we determined alveolar size using 3HeMR in healthy subjects aged 7 to 21 years. Alveolar dimensions did not increase by the expected rate over this age range, despite lung capacity increasing nearly fourfold. The only plausible explanation is new alveoli forming throughout the period of lung growth. Then, we compared alveolar size between children born very prematurely (<32 weeks gestation), including survivors of neonatal chronic lung disease (CLD) with term born children and children born mildly preterm (33-36 weeks gestation). Alveolar dimensions were nearly identical suggesting alveolar catch up growth in the very preterm groups. In the third part of the study, we investigated the relationship between various risk factors and alveolar dimensions. Exposure to environmental tobacco smoke (ETS) was found to be consistently associated with larger alveoli, suggesting its detrimental effect on alveolar development. Our results imply that developing lungs have the potential to recover from early life insults. Conversely, the window for adverse environmental exposures to affect alveolar development may be wider than previously believed.

No description available.

No description available.

In the Tg4 murine model of multiple sclerosis, experimental autoimmune encephalomyelitis, disease burden can be reduced by repeated doses of self-peptide, resulting in the conversion of pathogenic T H1 cells into tolerant IL-10-secreting regulatory cells. This work sought to identify whether there are any differences in the spatiotemporal organisation of signalling intermediates in these two cell types using live cell imaging of T cells transduced with GFP-tagged signalling intermediates. Pattern classification of distributions found at the T:B cell interface was used to identify any altered localisation. The distribution of all sensors tested have already been established in a model of foreign antigen (5C.C7 transgenic mouse) and the function of these interface structures inferred from the roles of proteins that congregate there. Thus we can relate structure to function and better understand the effects of altered localisation of proteins in tolerance.

No description available.

Interdisciplinary research plays an ever-increasing role in advancing the state-of-the art in bioimage analysis. Advances in biomedical imaging will give biologists the tools to analyse large numbers of electron and light microscopy images, quickly and accurately. Currently, many biologists analyse images manually. This is not reproducible and susceptible to subjective bias. Image segmentation and registration are two fundamental techniques, which are frequently applied to microscopy images. Cellular images contain many complex structures, and to be able to get accurate information from these images, standard segmentation and registration techniques cannot be applied. This thesis presents new. approaches to automatically analyse light and electron microscopy images. We first introduce a method to segment and measure insulin granule cores and membranes, from transmission electron microscopy images of beta cells of rat islets of Langerhans. The algorithm proceeds through two critical steps, firstly core segmentation and then membrane segmentation. For core segmentation we do an initial segmentation using a novel level-set active contour. Post-processing is done to remove over segmentation, from other organelles and particles within the cytoplasm. A final refining step is done using a novel dual level-set active contour. Membrane segmentation uses the initial segmentation, from core segmentation, as one of its inputs. Membranes are sampled and scaled, then gaps in the membrane are filled. It is possible that the granule membrane is at the core, so we introduce a novel convergence filter to verify our segmentation. We validate our method by doing comparisons against manual segmentation as well against the state-of-the-art. Our results also compare favourably to previously published data. We then propose a new algorithm for granule segmentation in 3-D transmission electron microscopy. Due to differences in sample preparation, the 3-D images have poor contrast. This presents a challenge for membrane segmentation. We propose a novel region-based active surface in a Bayesian framework with a core boundary prior, to overcome this. In some cases the granule membrane is not visible and is only characterized by a halo around the core. Our novel active surface shows promising results on our data set. Towards the end of the thesis we focus on a feature-based registration algorithm for largely misaligned bright-field light microscopy images and transmission electron microscopy images. We first detect cell centroids, using a gradient-based single-pass voting algorithm. Images are then aligned by finding the flip, translation and rotation parameters, which maximizes the overlap between pseudo-cell-centres. We demonstrate the effectiveness of our method, by comparing it to manually aligned images. Combining registered light and electron microscopy images together can reveal details about cellular structure with spatial and high-resolution information.

Macrophages are highly plastic cells of the innate immune system that play critical roles in immunity, development, homeostasis and tissue repair. However, they are also key mediators of inflammation and tissue damage in many diseases including autoimmunity. Experimental autoimmune uveoretinitis (EAU) is an animal model of human noninfectious posterior segment uveitis, and provides a useful platform to dissect the mechanisms of disease and trial new therapies pre-clinically. Retinal antigen-specific CD4+ T cells orchestrate EAU, and macrophages subsequently recruited into the eye become conditioned by T-cell cytokines including Th1-derived interferon (IFN)y and Th17-derived interleukin (IL)-17. Whilst IFNy is known to classically activate macrophages in a tumour necrosis factor (TNF)-dependent manner, the effect of IL-17 on macrophage activation is less well understood. A chronic low-level of inflammation has been shown to persist after the peak of disease in EAU, and this is associated with the development of pathological angiogenesis. Macrophages are also a key component of the wound healing machinery, and over the course of disease macrophage polarisation changes from a pro-inflammatory NOS2+ phenotype, to an Arg1+VEGP phenotype associated with tissue remodelling and angiogenesis. Therefore the aims of this thesis were to investigate how the mixed cytokine environment present during EAU influences macrophage activation, and how altering macrophage activation impacts the expression of disease during the chronic stages of EAU. The data presented here demonstrate that IL-17 alone has little impact on macrophage activation, but when combined with IFNy can synergistically increase the production of the tissue damaging molecule nitric oxide. Furthermore, altering macrophage activation through the abrogation of TNFR1 signalling results in a significant reduction in infiltrating leukocytes, and this is associated with a reduction in clinical disease score during the chronic phase of EAU. There is a marked change in the retinal cytokine environment in TNFR1 -/- mice, with significantly lower levels of TNF and IL-17. This was associated with a concomitant reduction in IL-17- producing T cells, but IFNy expression is maintained. At 90 days post immunisation, expression of angiogenesis in the retina is significantly reduced in TNFR1 -/- mice. TNF, but not IL-17, can directly promote angiogenesis through the activation of endothelial cells, and can therefore contribute to both inflammatory and angiogenic pathways, highlighting the pivotal role of TNF in all stages of EAU.