The diagnosis and treatment of cerebrovascular disease, such as stroke and vascular lesions in the brain, requires knowledge of the status of brain tissue and cerebral arteries. Perfusion imaging and angiography offer information on blood flow to the tissue and through the brain-feeding arteries. A variety of imaging modalities exist to assess brain haemodynamics, including measures of cerebral blood flow and other parameters, however many of these are invasive and/or involve the use of contrast agents toxic to certain patient populations. One non-invasive magnetic resonance imaging alternative for perfusion imaging and angiography, which also provides vessel specific blood flow information, is vessel-encoded pseudocontinuous arterial spin labelling (VEPCASL). VEPCASL uses the blood as an endogenous tracer and can magnetically label the blood from different arteries of interest. The combination of VEPCASL with different imaging strategies can provide a map of the vascular perfusion territories in the brain, or dynamic information on blood flow through the cerebral arteries. The work in this thesis seeks to optimise and develop the encoding process of VEPCASL and accompanying angiographic readouts. Firstly, a rapid and automated method for calculating a minimal number of signal-to-noise ratio (SNR) efficient encodings, for any number and arrangement of vessels, was developed. Its use resulted in improved SNR in perfusion territories fed by more complicated vessel arrangements in the labelling plane. The labelling efficiency of VEPCASL, and its non-vessel-selective equivalent, PCASL, is affected by the presence of magnetic field inhomogeneities in the labelling plane. Consequently, a correction for phase offsets was introduced into the calculation of the optimised encodings. These encodings enabled the recovery of SNR in perfusion territories for PCASL and VEPCASL when phase offsets were present at the labelled arteries. As current VEPCASL angiography methods are relatively slow to acquire, an accelerated readout was developed to acquire two-dimensional vessel-selective dynamic angiograms in approximately one minute. A radial k-space trajectory was found to offer the best vessel definition and SNR. Three-dimensional (3D) angiograms provide the most detailed view of the cerebral vasculature for use in diagnosis and treatment of cerebrovascular disease. A 3D radial readout was optimised to acquire vessel-selective dynamic angiograms. These angiograms offer information on the structure of the vascular tree and how it is fed by the major arteries in the neck. The techniques developed here aim to increase the clinical viability and applicability of VEPCASL perfusion imaging and angiography. It is hoped that the techniques herein could be used in patient populations to add to and improve the diagnostic information available.

The original description of the ischaemic penumbra asserted that both cerebral blood flow and metabolism would be required to monitor therapeutic intervention in acute ischaemic stroke. However, imaging in stroke trials has predominantly used biomarkers of infarction or perfusion-weighted signal to identify the pathophysiological processes that occur. This approach has neither identified novel treatment targets, nor been shown to consistently select patients who might benefit from intervention. The aim of this thesis was to use magnetic resonance imaging (MRI) biomarkers to identify and describe the pathophysiology of acute ischaemic stroke in patients. Patients admitted to the John Radcliffe Hospital in Oxford were recruited into a clinical imaging study. Serial imaging data were acquired, predominantly in the first hours after symptom onset, to capture the early dynamics of brain pathophysiology. Tissue status was meticulously defined over time to ensure robust interpretation of the novel imaging biomarkers: multiple post labeling delay arterial spin labeling to measure cerebral blood flow (CBF), and amide proton transfer to generate an intracellular pH-weighted signal. At a group level, the regions with the most severe injury had the lowest mean CBF and the greatest acidosis at presentation. There was a gradient of mean CBF and pH-weighted signal from the most ischaemic tissue to healthy tissue, but at the level of the individual there was considerable overlap in both parameters. The dynamics of perfusion were not sufficient to explain tissue outcome. Both acidosis and alkalosis were observed up to 24 hours in tissue that infarcted, and the nature of the pH change correlated with the timing of infarction. These data show that single imaging biomarkers cannot explain the pathophysiology of stroke and tissue fate. There is heterogeneity of pathophysiology both within and between patients, and the dynamics of these processes vary. Insight from pH-weighted imaging highlights the limitations of using perfusion imaging alone to assess tissue status, and supports the use of complementary metabolic imaging in the investigation of ischaemic stroke.

Ten people with multiple sclerosis (MS) who felt they had cognitive difficulties because of their MS were investigated. This study had multiple aims. Firstly, to explore the subjective experience of cognitive deficits. Secondly, to assess whether or not there was objective evidence of cognitive difficulties on neuropsychological testing, and whether this was commensurate with a pattern of subcortical dementia. Thirdly, to determine whether their magnetic resonance imaging (MRI) scans replicated the patterns of atrophy frequently reported in MS patients with cognitive difficulties. And finally, to investigate the psychological well-being of the subjects. In depth neuropsychiatric interviews, psychiatric and psychological inventories, a comprehensive neuropsychological battery, and MRI investigations were done. The mean Full Scale Intelligence Quotient (FSIQ) fell within the superior range, at the 89th percentile. On tests of general intelligence, mental state examinations, there was little or no indication of cognitive deterioration. However, on sophisticated neuropsychological testing, there was convincing evidence of cognitive problems. Magnetic resonance imaging lesions were atypical of the reported research on cognitively compromised MS patients.

M.Ing. / Automatic classification of images has always been an important part of pattern recognition. The segmentation and classification of MRI images has always been a challenge. A segmented image is often a very important input to the classification process. Many classification techniques use segmented images as input to the classification process. Certain segments or areas of an image serve as important features that will be used for classification. Important information can be derived from the features that are present in the segmented image. Sometimes there might be a need to extract a certain object from an image to do classification on the object. In the case of MRI images, certain structures of the human body like organs and tissue can be isolated by the segmentation process. These objects of interest (001) can give vital information for the identification of medical abnormalities (anomalies) and diseases. Segmented objects can play an important role to assist medical practitioners in the diagnosis and treatment of medical problems. I would like to test the performance of the watershed segmentation algorithm on MRI images of the cervical (C) spine. Much work has been done on the segmentation and classification of MRI images. Various techniques have been generated and tested over the past decades. Segmentation techniques like thresholding, convolution, pyramid segmentation and morphological segmentation have been utilised. All these techniques have their advantages and disadvantages. The pre-processing of an image plays a very important role in the success of the segmentation process. Histogram manipulation, filtering, thresholding and edge detection are important pre-processing techniques to yield good segmentation results. Many segmentation and classification techniques have been implemented on MRI images. The latest techniques include support vector machines (SVMs), neural networks (NNs), statistical methods, threshold techniques and normalised cuts. Segmentation of bony structures plays an important role in image guided surgery of the spine [1]. Physicians have commonly relied on computed tomography (CT) images to support their decisions in the diagnosis, treatment, and surgery of different pathologies of the spine due to the high resolution and good visualization of bone offered by this medical imaging modality. CT relies on the use of ionizing radiation, and does not depict soft tissue pathology, unlike magnetic resonance imaging (MRI) [1]. While the segmentation of vertebral bodies from CT images Segmentation Of C-Spine MRI Images Using The Watershed Transform Page 6 University of Johannesburg of the spine has commonly been accomplished with seed growing segmentation techniques [1], this task is more difficult in MRI, with variations in soft tissue contrast, and with the RF inhomogeneities, which increase the level of complexity. The primary goal of this project is to develop segmentation techniques for C-spine MRI images. This method will also be compared against other methods like pyramid segmentation and morphological segmentation. The watershed segmentation will be implemented and tested as the final step of the segmentation process. This project will try to use a combination of techniques, rather than to implement and evaluate one single method. It has been learned from literature and also from experience that the pre-processing of the raw data plays a crucial role in the quality of the segmentation process. Therefore, some attention will be given to the pre-processing of the images as part of the segmentation process.

T2 relaxation imaging can be used to measure signal from water trapped between myelin bilayers; the ratio of myelin water signal to total water is termed the myelin water fraction (MWF). First, results from multi-component T2 relaxation and diffusion tensor imaging (DTI) were compared for 19 multiple sclerosis (MS) subjects at 1.5 T to better understand how each measure is affected by pathology. In particular, it was determined that the detection of a long-T2 signal within an MS lesion may be indicative of a different underlying pathology than is present in lesions without long-T2 signal. Next, the single-slice T2 relaxation measurement was implemented, refined, and validated at 3.0 T. Scan parameters were varied for phantoms and in-vivo brain, and changes in multi-exponential fit residuals and T2 distribution-derived parameters such as MWF were monitored to determine which scan parameters minimized artifacts. Measurements were compared between 1.5 T and 3.0 T for 10 healthy volunteers. MWF maps were qualitatively similar between field strengths. MWFs were significantly higher at 3.0 T than at 1.5 T, but with a strong correlation between measurements at the different field strengths. Due to long acquisition times, multi-component T2 relaxation has thus far been clinically infeasible. The next study aimed to validate a new 3D multi-component T2 relaxation imaging technique against the 2D single-slice technique most commonly used. Ten healthy volunteers were scanned with both the 2D single-slice and 3D techniques. MWF maps were qualitatively similar between scans. MWF values were highly correlated between the acquisition methods. Although MWF values were generally lower using the 3D technique, they were only significantly so for peripheral brain structures, likely due to increased sensitivity of slab-selective refocusing pulses used for the 3D approach. The 3D T2 relaxation sequence was then applied to the study of MS to take advantage of the increased brain coverage. Thirteen MS subjects and 11 controls underwent T2 relaxation and DTI examinations to produce histograms of MWF and several DTI-derived metrics. MS MWF histograms differed considerably from those of controls, and differences in MS MWF histograms did not mirror differences in DTI histograms relative to matched controls. / Science, Faculty of / Physics and Astronomy, Department of / Graduate

Structural brain networks can be constructed at a macroscopic scale using diffusion magnetic resonance imaging (dMRI) and whole-brain tractography. Under this approach, grey matter regions, such as Brodmann areas, form the nodes of a network and tractography is used to construct a set of white matter fibre tracts which form the connections. Graph-theoretic measures may then be used to characterise patterns of connectivity. In this study, we measured the test-retest properties of such networks by varying several factors affecting network construction using ten healthy volunteers who underwent a dMRI protocol at 1.5 T on two separate occasions. High resolution T1-weighted brains were parcellated into regions-of-interest and network connections were identified using dMRI and two alternative tractography algorithms, two alternative seeding strategies, constraints on anatomical plausibility and three alternative network weightings. Test-retest performance was found to improve when: 1) seeding from white matter, rather than grey; and 2) using probabilistic tractography, rather than deterministic. In terms of network weighting, a measure of streamline density produced better test-retest performance than tract-averaged diffusion anisotropy, although it remains unclear which is most representative of the underlying axonal connections. These findings were then used to inform network construction for two further cohorts: a casecontrol analysis of 30 patients with amyotrophic lateral sclerosis (ALS) compared with 30 age-matched healthy controls; and a cross-sectional analysis of 80 healthy volunteers aged 25– 64 years. In both cases, networks were constructed using a weighting reflecting tract-averaged fractional anisotropy (FA). A mass-univariate statistical technique called network-based statistics, identified an impaired motor-frontal-subcortical subnetwork (10 nodes and 12 bidirectional connections), consistent with upper motor neuron pathology, in the ALS group compared with the controls. Reduced FA for three of the impaired network connections, which involved fibres of the cortico-spinal tract, were significantly correlated with the rate of disease progression. Cross-sectional analysis of the 80 healthy volunteers was intended to provide supporting evidence for the widely reported age-related decline in white matter integrity. However, no meaningful relationships were found between increasing age and impaired connectivity based on global, lobar and nodal network properties – findings which were confirmed with a conventional voxel-based analysis of the dMRI data. In conclusion, whilst current acquisition protocols and methods can produce networks capable of characterising the genuine between-subject differences in connectivity, it is challenging to measure subtle white matter changes, for example, due to normal ageing. We conclude that future work should be undertaken to address these concerns.

A novel framework for quantitative analysis of shape and function in magnetic resonance imaging (MRI) of the brain is proposed. First, an efficient method to compute invariant spherical harmonics (SPHARM) based feature representation for real valued 3D functions was developed. This method addressed previous limitations of obtaining unique feature representations using a radial transform. The scale, rotation and translation invariance of these features enables direct comparisons across subjects. This eliminates need for spatial normalization or manually placed landmarks required in most conventional methods [1-6], thereby simplifying the analysis procedure while avoiding potential errors due to misregistration. The proposed approach was tested on synthetic data to evaluate its improved sensitivity. Application on real clinical data showed that this method was able to detect clinically relevant shape changes in the thalami and brain ventricles of Parkinson's disease patients. This framework was then extended to generate functional features that characterize 3D spatial activation patterns within ROIs in functional magnetic resonance imaging (fMRI). To tackle the issue of intersubject structural variability while performing group studies in functional data, current state-of-the-art methods use spatial normalization techniques to warp the brain to a common atlas, a practice criticized for its accuracy and reliability, especially when pathological or aged brains are involved [7-11]. To circumvent these issues, a novel principal component subspace was developed to reduce the influence of anatomical variations on the functional features. Synthetic data tests demonstrate the improved sensitivity of this approach over the conventional normalization approach in the presence of intersubject variability. Furthermore, application to real fMRI data collected from Parkinson's disease patients revealed significant differences in patterns of activation in regions undetected by conventional means. This heightened sensitivity of the proposed features would be very beneficial in performing group analysis in functional data, since potential false negatives can significantly alter the medical inference. The proposed framework for reducing effects of intersubject anatomical variations is not limited to functional analysis and can be extended to any quantitative observation in ROIs such as diffusion anisotropy in diffusion tensor imaging (DTI), thus providing researchers with a robust alternative to the controversial normalization approach. / Applied Science, Faculty of / Electrical and Computer Engineering, Department of / Graduate

Multi-echo T2 relaxation measurements of the human spinal cord (SC) reveal a short T2 pool of water believed to arise from water trapped between myelin bilayers, where the proportion of this water to the total water signal is called the myelin water fraction (MWF). In the present study, MWF were measured in the healthy human cervical spine at the C4-C6 vertebral levels in vivo using a 3D modified 32 echo CPMG sequence to acquire axial slices perpendicular to the cord. Volunteers were recruited in two age ranges, under 30 years old and over 50 years old, and a subset of both groups were scanned twice to test reproducibility. Mean MWF in the dorsal and lateral column WM of the group under 30 years of age was 0.29 (0.01) (mean(SE)), which agrees with previously reported MWF values in the cervical spine. The mean absolute difference between two scans was 0.06 or 26%. A negative correlation between WM MWF and age was hinted at in these findings, however more subjects are required to improve statistical power. This study paves the way for the use of 3D myelin water imaging in the cervical spine at 3.0T for the assessment of SC WM pathology. / Science, Faculty of / Physics and Astronomy, Department of / Graduate

The recovery of mineral from ore in the heap leaching process requires the mineral to come into contact with the leaching solution. However, heap hydrology is prone to non-uniform behaviour due to the inhomogeneity of the ore particles and the heap structure. The primary aim of this thesis was therefore to develop a magnetic resonance imaging (MRI) technique which would allow for the novel non-invasive imaging of the liquid hold-up in representative laboratory-scale heap leaching systems. The ferro- and paramagnetic species in the ore were found to cause significant distortions in frequency encoded MRI acquisitions. These distortions were mitigated through the application of single point imaging techniques. Comparison with equivalent X-ray CT acquisitions, which are immune to magnetic susceptibility distortions, confirmed the accuracy of the single point acquisitions. Spin echo single point imaging (SESPI) was demonstrated to be more robust than the simple single point imaging (SPI) technique because the effect of $\textit{T}$$_{2}$ weighting on the MRI signal was less significant than that of $\textit{T}$$_{2}$*. Mapping of the gas, liquid and solid distributions in unsaturated leaching columns using SESPI allowed for the quantification of the hold-up and voidage values. These compared favourably to gravimetric measurements and literature values, thereby confirming their validity. Novel measurement of the liquid film thickness and the interfacial areas between the three phases was achieved and insight into the liquid behaviour was obtained through different flow rate experiments as well as MRI tracer experiments. Drip irrigation of the ore was considered for beds of large, fine and agglomerated ore. Liquid distribution in the fine ore closely resembled that of irrigation of clay soils whereas flow through the large ore was gravity dominated. Slumping of the agglomerated ore caused permanent limitation of the liquid distribution and the effect of flow rate changes was observed to be limited to the region below the irrigation point. In both the column and drip irrigation experiments, increases in the liquid flow rate caused small increases in the liquid hold-up in the form of new, relatively thicker rivulets, thereby improving the liquid-solid interfacial area. However it did not affect capillary held liquid volume so the effect of flow rate changes decreased as the lateral distance from the irrigation point increased. Finally, X-ray CT images of a long term ferric leach were analysed to assess the factors affecting the mineral leaching. Proximity of the mineral to the ore surface was found to be the critical factor, as only mineral located within 2 mm of the surface was recovered, though liquid distribution effects were also observed.

Recent reviews (Peyser & Poser, 1986; Rao, 1986) suggest that Multiple Sclerosis results in cognitive impairments in the areas of learning and memory, abstract reasoning, information processing efficiency, and, often, visual-spatial ability. Whether this pattern applies to the individual with MS is unclear. Due to the disseminated distribution of MS neuropathology, patients may undergo idiosyncratic cognitive changes dependent upon the site of white matter lesions. The present study explored this question using cluster analysis on the neuropsychological data from a group of mildly disabled MS patients (n = 177) and a well-matched control group (n=89). In a group of MS patients who were identified with unequivocal cognitive impairment, the resultant clusters indicated that MS does not result in a characteristic pattern of impairment. Two clusters were obtained that resembled the pattern described in the literature, while the majority of patients clustered into groups with specific deficits in one or two areas, with normal performance in others. In order to identify associations between cluster groups and lesion sites, frequency tables were constructed for the presence of a lesion on Magnetic Resonance Imaging in 24 brain sites. An association was obtained between two lesion sites and two cognitive tests. Visual-spatial impairment, as assessed by the Benton Visual Retention test, was associated with lesions in the genu of the corpus callosum and with more lesions throughout the corpus callosum. Impaired performance on Paired Associates, a test of learning and memory for novel verbal associations, was associated with a lesion in the deep white matter of the left parietal lobe. The results support the hypothesis that MS results in multiple patterns of cognitive impairment depending on the individual placement of white matter lesions. Identifying and characterizing the heterogeneity of the impairment may greatly increase our understanding of the role of myelin in cognition and the functions of white matter tracts in the brain. / Arts, Faculty of / Psychology, Department of / Graduate