Advances in medical imaging offer valuable tools towards a better understanding of the human anatomy and the diversity of physiological and pathological processes present in the body. Image segmentation is a crucial building block in medical image analysis, allowing quantitative analysis of the image content. It is therefore used as a fundamental step in a myriad of clinical applications. As the wealth of available data increases, automating this process is imperative as manual segmentation is very time consuming and may require high levels of expertise. Automated segmentation approaches face challenges in large databases due to large variability in shape and appearance of the structures of interest, the presence of pathologies, or different imaging protocols used to acquire the images. In particular, it becomes increasingly desirable to develop robust and accurate segmentation techniques that rely on minimal manual input or weak supervision. The contributions presented in this thesis are three-fold: We present two graphical methods for atlas-based anatomical image segmentation with a focus on alleviating the impact of limited availability of manually annotated training data. In addition, we present a probabilistic anatomical model based on segmentation which can help analyse regional morphological differences between groups, e.g. between healthy subjects and dementia patients, and provide local disease scores in the images which express the estimated degree of pathology. The proposed segmentation methods were evaluated on segmentation tasks on fetal and adult brain MR images and adult cardiac MR images. Experiments demonstrated the flexibility of the proposed approaches and showed that accurate segmentation results could be obtained while reducing the amount of manually annotated input. The proposed anatomical disease model was evaluated on a clinical brain MR dataset for differentiation between Alzheimer’s disease patients and healthy elderly subjects. Experiments showed that modelling disease-specific anatomy in higher detail revealed a greater sensitivity to local anatomical differences between the studied groups.

Cone beam computed tomography (CBCT) enables a volumetric image reconstruction from a set of 2D projection data. This thesis studies the performance of a wide range iterative algorithms in various aspects, aiming to generate a better CBCT image reconstruction, especially when projection data is limited. We have implemented a wide range of algebraic iterative algorithms. Hence, the performance of ART, SART and OS-SART is studied based on a range of image quality measures. The major limitations of traditional iterative methods are their computational time. The conjugate gradients (CG) algorithm and its variants can be used to solve linear systems of equations arising from CBCT. Their applications can be found in a general linear algebra context, but in tomography problems (e.g. CBCT reconstruction) they have not widely been used. Hence, CBCT reconstruction using the CG-type algorithm LSQR was implemented and studied. In CBCT reconstruction, the main computational challenge is that the matrix A usually is very large, and storing it in full requires an amount of memory well beyond the reach of commodity computers. Because of these memory capacity constraints, only a small fraction of the weighting matrix A is typically used, leading to a poor reconstruction. In this final part of the thesis, to overcome this diculty, the matrix A is partitioned and stored blockwise, and blockwise matrix-vector multiplications are implemented within LSQR. This implementation allows us to use the full weighting matrix A for CBCT reconstruction without further enhancing computer standards. Tikhonov regularization has been developed in this framework, and can produce significant improvement in the reconstructed images for limited data case.

Electrical impedance tomography (EIT) is an imaging technique for detecting the internal conductivity distribution of an object by voltage measurements taken by an exterior electrode. EIT has been researched in many different application areas in the world as a simpler, cheaper alternative to many other imaging methods. The topic of this PhD study is mainly focused on a number of key developments in both hardware and software implementation. The basic theories of EIT, including forward problem, inverse problem of EIT and the sensor design have been described. Major contributions of the thesis are in computational and experimental aspects of EIT in a wide variety of geometries. A sparse and memory efficient method has been presented to solve large scale 3D EIT problems. A parallel conjugate gradient (PCG) has been applied to demonstrate computational improvements using synthetic and experimental data. 3D EIT has been implemented for planar array geometry for limited access tomography. Furthermore, multiple frequencies with complex conductivity reconstruction are presented and applied to an EIT-based fabric pressure mapping sensor. A comparative study with traditional tank phantom is presented to provide a context for a fabric pressure mapping sensor. As the motivation for different frequency response with different conductivity inclusions, frequency difference EIT has been implemented to overcome problems of time difference EIT.

The ZFP36 proteins are a family of post-transcriptional regulator proteins that bind to adenine uridine rich elements (AREs) in 3’ untranslated (3’UTR) regions of mRNAs. The members of the human family, ZFP36L1, ZFP36L2 and ZFP36 are able to degrade mRNAs of important cell regulators that include cytokines, cell signalling proteins and transcriptional factors. This project investigated two proposed targets for the protein family that have important roles in B cell biology, BCL2 and CD38 mRNAs. BCL2 is an anti-apoptotic protein with key roles in cell survival and carcinogenesis; CD38 is a membrane protein differentially expressed in B cells and with a prognostic value in B chronic lymphocytic leukaemia (B-CLL), patients positive for CD38 are considered to have a poor prognosis. This project provides evidence of a functional interaction between the three ZFP36 proteins and the 3’UTR AREs of BCL2 and CD38 mRNAs. 3’UTR dual luciferase reporter assay results showed that the three ZFP36 proteins bound the 3’UTR ARE of BCL2 mRNA and CD38 mRNA. Zinc finger mutant versions of ZFP36L1 failed to bind the 3’ UTR AREs for each target, proving that intact zinc finger domains are the functional binding domains of the protein and are required for interaction with AREs. A complete ARE sequence is also needed and when mutated BCL2 3’UTR ARE was tested, lacking the adenine uridine rich core element, the BCL2 transcript was not bound by ZFP36L1 protein. For CD38 further experiments have demonstrated that down regulation of ZFP36L1 by siRNAs in HeLa cells resulted in an increase in CD38 expression as measured by immunofluorescence and flow cytometry and by Western blot analysis. These results provide further evidence that ZFP36L1 negatively regulates CD38 mRNA. Analysis of BCL2, CD38 and ZFP36L1 protein expression in primary B-CLL cells by Western blot analysis did not show an inverse relationship between the proposed targets and ZFP36L1. Protein expression analysis in B-CLL for the whole family of ZFP36 proteins showed that ZFP36L1 was heterogeneously expressed; ZFP36L2 was detected at very low levels or was undetectable and ZFP36 was low and homogeneously expressed. In cell lines representing different B cell stages, but mainly representing mature and plasma cell stages, ZFP36L2 was detected at relatively high levels but also heterogeneously and there was very low or undetectable expression of ZFP36L1 in all cells. Immunohistochemistry analysis of ZFP36L1, BCL2 and CD38 in normal lymphoid tissue and FL indicated that areas of normal lymphoid tissues associated with highest levels of BCL2 and CD38 were associated with low or undetectable levels of ZFP36L1. In FL (FL) ZFP36L1 was detected in follicular centre cells, where BCL2 is also reported to be highly expressed due to a translocation that leads to over expression of BCL2. CD38 expression was also detected within FL follicle centres with some cells showing a high level of expression within the neoplastic follicle and amongst scattered cells outside of it. Overall, the results support the hypothesis that ZFP36L1 (and also ZFP36 and ZFP36L2) negatively regulates BCL2 and CD38 mRNAs. In a wider context, the results of this project support the view that ZFP36L1, and perhaps other ZFP36 family proteins, play important roles in controlling mature B cell survival and differentiation by targeting important regulatory mRNAs in these cells.

This thesis describes the value of the free-response receiver operating characteristic (FROC) paradigm for dose and image quality optimisation in a niche area of imaging. The empirical works discussed in this thesis focus on the diagnostic value of the low-resolution computed tomography (CT) images acquired for attenuation correction (AC) – a process primarily used to correct for photon attenuation with images produced merely as a consequence of the exposure. The potential discovery of incidental findings on these images was investigated. The observers taking part in the empirical studies were generally lacking in significant experience of interpreting CT images. As a consequence it was also deemed valuable to investigate the value of the novice observer in free-response studies. A further methodological consideration for studies of this kind is consistent and reliable image display and FROC data collection. Prototype software, ROCView, was designed and developed to make this an easy process and the key functionality and impact is analysed here. In addition to the empirical works, two review papers, aimed at the technologists and radiographers performing low-resolution CT for AC, are summarised. They explain the value of the FROC paradigm and the jackkinfe alternative FROC (JAFROC) analysis method to a wide audience in nuclear medicine.

Surfactant protein D (SP-D) is a carbohydrate/charged pattern recognition molecule of the innate immune system. By virtue of its ability to recognize an array of carbohydrate patterns on the surface of a range of pathogens, SP-D can bring about opsonisation, enhanced phagocytosis and killing of a diverse range of viruses, bacteria and fungi. In addition to antimicrobial functions, which also includes bacteriostatic and fungistatic properties SP-D has also been shown to bind allergens derived from a number of sources including house dust mite, Aspergilllus fumigatus and pollen grains. SP-D allergen interaction leads to inhibition of specific IgE binding and subsequent downregulation of histamine release from sensitized basophils and mast cells. Thus, a number of murine models of pulmonary hypersensitivity and allergic asthma induced by ovalbumin, house dust mite and Aspergillus fumigatus allergens/antigens have been tested for the ability of SP-D to dampen allergic symptoms on the immunological parameters. In general, treatment of allergic models with a recombinant fragment of human SP-D (rh SP-D; composed of trimeric, neck and carbohydrate recognition domain) has been shown to cause downregualtion of specific IgE synthesis, pulmonary and peripheral eosinophilia and airway hyper reactivity, and Th2→Th1 polarisation. However, therapeutic alleviation of eosinophilia by rh SP-D treatment became evident when SP-D gene deficient mice were found to be hypereosinophilic In fact, rhSP-D binds well to eosinophils derived from allergic patients and induces apoptosis without affecting eosinophils derived from healthy individuals or non-activated/non-sensitized eosinophils. Proteomic analysis of rh SP-D treated eosinophillic cell line that revealed that apoptosis induction takes place via p53 pathway. In this thesis, proteomic signatures were replicated using a leukemic cell line AML14.3D10 via qPCR analysis by identifying targets from a spectrum of genes, which were either upregulated or downregulated. It appears that in spite of induction of apoptosis by rh SP-D, different cells respond differentially at molecular levels (Chapter 3). Sensing that SP-D can induce apoptosis in altered or transformed cells; the effect of SP-D gene expression within pancreatic cancer cells was also investigated. The experiments confirmed p53 pathway dependence for suppression of cancer. Interestingly, factors responsible for metastasis for cancer are also downregulated by endogenous overexpression of SP-D, as validated by wound healing assay. We conclude that SP-D is a general immunosurveillance molecule, which is involved in the clearance of altered and transformed cells (Chapter 4). Chapter 5 shows a direct interaction between DC-SIGN and rh SP-D that inhibits DC-SIGN interaction of allergens and HIV-1, tow common ligands for SP-D and DC-SIGN. Using transfected human embryonic kidney (HEK) cells expressing surface DC-SIGN, we found that pre-treatment of these cells with rhSP-D suppressed DC-SIGN mediated transmission of HIV-1to co-cultured PBMCs. The effect of rhSP-D-DC-SIGN In conclusion, this thesis highlights a broader immune role of SP-D in homeostasis and probably assigns potential functions of extrapulmonary and/or locally synthesized SP-D within non-lung tissues and blood.

No description available.

No description available.

No description available.

Camel heavy chain only antibodies have evolved into a strikingly different structural arrangement compared to conventional antibodies. Heavy chain-only antibodies do not interact with a light chain, they lack a CHI domain and are regularly observed to have an extended CDR-H3 region. Due to the presence of both antibody types in the camel species, a unique opportunity is available to compare their structure/activity relationship in a single immune response. This thesis aimed to isolate VHH (antigen binding domain of heavy chain-only antibodies) and conventional camel antibody Fab fragments, to assess whether they bind to different structural features of a cytokine. Further work aimed to assess whether VHH fragments were capable of stabilising a putative cleft of the cytokine and neutralise signalling. A fusion protein of the cytokine and its alpha receptor was cloned, purified, and used in camel immunisations to generate an immune response. A method was developed that enabled isolation of antibody cDNA from individual antigen positive camel B-cells, which in turn yielded a panel of 38 specific camel heavy chain-only and two conventional antibodies. Following expression and purification, the antibodies’ activities were profiled using Biacore, ELISA and functional cell-based assays to map the expected sites of interaction on the cytokine. These sites of interaction were structurally confirmed using a rapid NMR epitope mapping technique. Crystal structures were achieved for five VHH fragments and one conventional Fab fragment in complex with the cytokine. The interaction of conventional antibodies with proteins is associated with large planar surfaces. Limited studies undertaken to examine the binding interactions of camelid VHH fragments have shown them to be able to bind to clefts on a protein surface. Data generated in this thesis show that both camel VHH and Fab fragments display a cleft-binding nature, with the former also able to interact with planar surfaces. Both Fab and VHH fragments displayed neutralising activity, with the cleft-binding VHH fragments stabilising the structure of the cytokine.